AI is developing all the time. Here are some picks from several articles what is expected to happen in AI and around it in 2025. Here are picks from various articles, the texts are picks from the article edited and in some cases translated for clarity.
AI in 2025: Five Defining Themes
https://news.sap.com/2025/01/ai-in-2025-defining-themes/
Artificial intelligence (AI) is accelerating at an astonishing pace, quickly moving from emerging technologies to impacting how businesses run. From building AI agents to interacting with technology in ways that feel more like a natural conversation, AI technologies are poised to transform how we work.
But what exactly lies ahead?
1. Agentic AI: Goodbye Agent Washing, Welcome Multi-Agent Systems
AI agents are currently in their infancy. While many software vendors are releasing and labeling the first “AI agents” based on simple conversational document search, advanced AI agents that will be able to plan, reason, use tools, collaborate with humans and other agents, and iteratively reflect on progress until they achieve their objective are on the horizon. The year 2025 will see them rapidly evolve and act more autonomously. More specifically, 2025 will see AI agents deployed more readily “under the hood,” driving complex agentic workflows.
In short, AI will handle mundane, high-volume tasks while the value of human judgement, creativity, and quality outcomes will increase.
2. Models: No Context, No Value
Large language models (LLMs) will continue to become a commodity for vanilla generative AI tasks, a trend that has already started. LLMs are drawing on an increasingly tapped pool of public data scraped from the internet. This will only worsen, and companies must learn to adapt their models to unique, content-rich data sources.
We will also see a greater variety of foundation models that fulfill different purposes. Take, for example, physics-informed neural networks (PINNs), which generate outcomes based on predictions grounded in physical reality or robotics. PINNs are set to gain more importance in the job market because they will enable autonomous robots to navigate and execute tasks in the real world.
Models will increasingly become more multimodal, meaning an AI system can process information from various input types.
3. Adoption: From Buzz to Business
While 2024 was all about introducing AI use cases and their value for organizations and individuals alike, 2025 will see the industry’s unprecedented adoption of AI specifically for businesses. More people will understand when and how to use AI, and the technology will mature to the point where it can deal with critical business issues such as managing multi-national complexities. Many companies will also gain practical experience working for the first time through issues like AI-specific legal and data privacy terms (compared to when companies started moving to the cloud 10 years ago), building the foundation for applying the technology to business processes.
4. User Experience: AI Is Becoming the New UI
AI’s next frontier is seamlessly unifying people, data, and processes to amplify business outcomes. In 2025, we will see increased adoption of AI across the workforce as people discover the benefits of humans plus AI.
This means disrupting the classical user experience from system-led interactions to intent-based, people-led conversations with AI acting in the background. AI copilots will become the new UI for engaging with a system, making software more accessible and easier for people. AI won’t be limited to one app; it might even replace them one day. With AI, frontend, backend, browser, and apps are blurring. This is like giving your AI “arms, legs, and eyes.”
5. Regulation: Innovate, Then Regulate
It’s fair to say that governments worldwide are struggling to keep pace with the rapid advancements in AI technology and to develop meaningful regulatory frameworks that set appropriate guardrails for AI without compromising innovation.
12 AI predictions for 2025
This year we’ve seen AI move from pilots into production use cases. In 2025, they’ll expand into fully-scaled, enterprise-wide deployments.
https://www.cio.com/article/3630070/12-ai-predictions-for-2025.html
This year we’ve seen AI move from pilots into production use cases. In 2025, they’ll expand into fully-scaled, enterprise-wide deployments.
1. Small language models and edge computing
Most of the attention this year and last has been on the big language models — specifically on ChatGPT in its various permutations, as well as competitors like Anthropic’s Claude and Meta’s Llama models. But for many business use cases, LLMs are overkill and are too expensive, and too slow, for practical use.
“Looking ahead to 2025, I expect small language models, specifically custom models, to become a more common solution for many businesses,”
2. AI will approach human reasoning ability
In mid-September, OpenAI released a new series of models that thinks through problems much like a person would, it claims. The company says it can achieve PhD-level performance in challenging benchmark tests in physics, chemistry, and biology. For example, the previous best model, GPT-4o, could only solve 13% of the problems on the International Mathematics Olympiad, while the new reasoning model solved 83%.
If AI can reason better, then it will make it possible for AI agents to understand our intent, translate that into a series of steps, and do things on our behalf, says Gartner analyst Arun Chandrasekaran. “Reasoning also helps us use AI as more of a decision support system,”
3. Massive growth in proven use cases
This year, we’ve seen some use cases proven to have ROI, says Monteiro. In 2025, those use cases will see massive adoption, especially if the AI technology is integrated into the software platforms that companies are already using, making it very simple to adopt.
“The fields of customer service, marketing, and customer development are going to see massive adoption,”
4. The evolution of agile development
The agile manifesto was released in 2001 and, since then, the development philosophy has steadily gained over the previous waterfall style of software development.
“For the last 15 years or so, it’s been the de-facto standard for how modern software development works,”
5. Increased regulation
At the end of September, California governor Gavin Newsom signed a law requiring gen AI developers to disclose the data they used to train their systems, which applies to developers who make gen AI systems publicly available to Californians. Developers must comply by the start of 2026.
There are also regulations about the use of deep fakes, facial recognition, and more. The most comprehensive law, the EU’s AI Act, which went into effect last summer, is also something that companies will have to comply with starting in mid-2026, so, again, 2025 is the year when they will need to get ready.
6. AI will become accessible and ubiquitous
With gen AI, people are still at the stage of trying to figure out what gen AI is, how it works, and how to use it.
“There’s going to be a lot less of that,” he says. But gen AI will become ubiquitous and seamlessly woven into workflows, the way the internet is today.
7. Agents will begin replacing services
Software has evolved from big, monolithic systems running on mainframes, to desktop apps, to distributed, service-based architectures, web applications, and mobile apps. Now, it will evolve again, says Malhotra. “Agents are the next phase,” he says. Agents can be more loosely coupled than services, making these architectures more flexible, resilient and smart. And that will bring with it a completely new stack of tools and development processes.
8. The rise of agentic assistants
In addition to agents replacing software components, we’ll also see the rise of agentic assistants, adds Malhotra. Take for example that task of keeping up with regulations.
Today, consultants get continuing education to stay abreast of new laws, or reach out to colleagues who are already experts in them. It takes time for the new knowledge to disseminate and be fully absorbed by employees.
“But an AI agent can be instantly updated to ensure that all our work is compliant with the new laws,” says Malhotra. “This isn’t science fiction.”
9. Multi-agent systems
Sure, AI agents are interesting. But things are going to get really interesting when agents start talking to each other, says Babak Hodjat, CTO of AI at Cognizant. It won’t happen overnight, of course, and companies will need to be careful that these agentic systems don’t go off the rails.
Companies such as Sailes and Salesforce are already developing multi-agent workflows.
10. Multi-modal AI
Humans and the companies we build are multi-modal. We read and write text, we speak and listen, we see and we draw. And we do all these things through time, so we understand that some things come before other things. Today’s AI models are, for the most part, fragmentary. One can create images, another can only handle text, and some recent ones can understand or produce video.
11. Multi-model routing
Not to be confused with multi-modal AI, multi-modal routing is when companies use more than one LLM to power their gen AI applications. Different AI models are better at different things, and some are cheaper than others, or have lower latency. And then there’s the matter of having all your eggs in one basket.
“A number of CIOs I’ve spoken with recently are thinking about the old ERP days of vendor lock,” says Brett Barton, global AI practice leader at Unisys. “And it’s top of mind for many as they look at their application portfolio, specifically as it relates to cloud and AI capabilities.”
Diversifying away from using just a single model for all use cases means a company is less dependent on any one provider and can be more flexible as circumstances change.
12. Mass customization of enterprise software
Today, only the largest companies, with the deepest pockets, get to have custom software developed specifically for them. It’s just not economically feasible to build large systems for small use cases.
“Right now, people are all using the same version of Teams or Slack or what have you,” says Ernst & Young’s Malhotra. “Microsoft can’t make a custom version just for me.” But once AI begins to accelerate the speed of software development while reducing costs, it starts to become much more feasible.
9 IT resolutions for 2025
https://www.cio.com/article/3629833/9-it-resolutions-for-2025.html
1. Innovate
“We’re embracing innovation,”
2. Double down on harnessing the power of AI
Not surprisingly, getting more out of AI is top of mind for many CIOs.
“I am excited about the potential of generative AI, particularly in the security space,”
3. And ensure effective and secure AI rollouts
“AI is everywhere, and while its benefits are extensive, implementing it effectively across a corporation presents challenges. Balancing the rollout with proper training, adoption, and careful measurement of costs and benefits is essential, particularly while securing company assets in tandem,”
4. Focus on responsible AI
The possibilities of AI grow by the day — but so do the risks.
“My resolution is to mature in our execution of responsible AI,”
“AI is the new gold and in order to truly maximize it’s potential, we must first have the proper guardrails in place. Taking a human-first approach to AI will help ensure our state can maintain ethics while taking advantage of the new AI innovations.”
5. Deliver value from generative AI
As organizations move from experimenting and testing generative AI use cases, they’re looking for gen AI to deliver real business value.
“As we go into 2025, we’ll continue to see the evolution of gen AI. But it’s no longer about just standing it up. It’s more about optimizing and maximizing the value we’re getting out of gen AI,”
6. Empower global talent
Although harnessing AI is a top objective for Morgan Stanley’s Wetmur, she says she’s equally committed to harnessing the power of people.
7. Create a wholistic learning culture
Wetmur has another talent-related objective: to create a learning culture — not just in her own department but across all divisions.
8. Deliver better digital experiences
Deltek’s Cilsick has her sights set on improving her company’s digital employee experience, believing that a better DEX will yield benefits in multiple ways.
Cilsick says she first wants to bring in new technologies and automation to “make things as easy as possible,” mirroring the digital experiences most workers have when using consumer technologies.
“It’s really about leveraging tech to make sure [employees] are more efficient and productive,”
“In 2025 my primary focus as CIO will be on transforming operational efficiency, maximizing business productivity, and enhancing employee experiences,”
9. Position the company for long-term success
Lieberman wants to look beyond 2025, saying another resolution for the year is “to develop a longer-term view of our technology roadmap so that we can strategically decide where to invest our resources.”
“My resolutions for 2025 reflect the evolving needs of our organization, the opportunities presented by AI and emerging technologies, and the necessity to balance innovation with operational efficiency,”
Lieberman aims to develop AI capabilities to automate routine tasks.
“Bots will handle common inquiries ranging from sales account summaries to HR benefits, reducing response times and freeing up resources for strategic initiatives,”
Not just hype — here are real-world use cases for AI agents
https://venturebeat.com/ai/not-just-hype-here-are-real-world-use-cases-for-ai-agents/
Just seven or eight months ago, when a customer called in to or emailed Baca Systems with a service question, a human agent handling the query would begin searching for similar cases in the system and analyzing technical documents.
This process would take roughly five to seven minutes; then the agent could offer the “first meaningful response” and finally begin troubleshooting.
But now, with AI agents powered by Salesforce, that time has been shortened to as few as five to 10 seconds.
Now, instead of having to sift through databases for previous customer calls and similar cases, human reps can ask the AI agent to find the relevant information. The AI runs in the background and allows humans to respond right away, Russo noted.
AI can serve as a sales development representative (SDR) to send out general inquires and emails, have a back-and-forth dialogue, then pass the prospect to a member of the sales team, Russo explained.
But once the company implements Salesforce’s Agentforce, a customer needing to modify an order will be able to communicate their needs with AI in natural language, and the AI agent will automatically make adjustments. When more complex issues come up — such as a reconfiguration of an order or an all-out venue change — the AI agent will quickly push the matter up to a human rep.
Open Source in 2025: Strap In, Disruption Straight Ahead
Look for new tensions to arise in the New Year over licensing, the open source AI definition, security and compliance, and how to pay volunteer maintainers.
https://thenewstack.io/open-source-in-2025-strap-in-disruption-straight-ahead/
The trend of widely used open source software moving to more restrictive licensing isn’t new.
In addition to the demands of late-stage capitalism and impatient investors in companies built on open source tools, other outside factors are pressuring the open source world. There’s the promise/threat of generative AI, for instance. Or the shifting geopolitical landscape, which brings new security concerns and governance regulations.
What’s ahead for open source in 2025?
More Consolidation, More Licensing Changes
The Open Source AI Debate: Just Getting Started
Security and Compliance Concerns Will Rise
Paying Maintainers: More Cash, Creativity Needed
Kyberturvallisuuden ja tekoälyn tärkeimmät trendit 2025
https://www.uusiteknologia.fi/2024/11/20/kyberturvallisuuden-ja-tekoalyn-tarkeimmat-trendit-2025/
1. Cyber infrastructure will be centered on a single, unified security platform
2. Big data will give an edge against new entrants
3. AI’s integrated role in 2025 means building trust, governance engagement, and a new kind of leadership
4. Businesses will adopt secure enterprise browsers more widely
5. AI’s energy implications will be more widely recognized in 2025
6. Quantum realities will become clearer in 2025
7. Security and marketing leaders will work more closely together
Presentation: For 2025, ‘AI eats the world’.
https://www.ben-evans.com/presentations
Just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity.
https://www.securityweek.com/ai-implementing-the-right-technology-for-the-right-use-case/
If 2023 and 2024 were the years of exploration, hype and excitement around AI, 2025 (and 2026) will be the year(s) that organizations start to focus on specific use cases for the most productive implementations of AI and, more importantly, to understand how to implement guardrails and governance so that it is viewed as less of a risk by security teams and more of a benefit to the organization.
Businesses are developing applications that add Large Language Model (LLM) capabilities to provide superior functionality and advanced personalization
Employees are using third party GenAI tools for research and productivity purposes
Developers are leveraging AI-powered code assistants to code faster and meet challenging production deadlines
Companies are building their own LLMs for internal use cases and commercial purposes.
AI is still maturing
However, just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity. Right now, we very much see AI in this “peak of inflated expectations” phase and predict that it will dip into the “trough of disillusionment”, where organizations realize that it is not the silver bullet they thought it would be. In fact, there are already signs of cynicism as decision-makers are bombarded with marketing messages from vendors and struggle to discern what is a genuine use case and what is not relevant for their organization.
There is also regulation that will come into force, such as the EU AI Act, which is a comprehensive legal framework that sets out rules for the development and use of AI.
AI certainly won’t solve every problem, and it should be used like automation, as part of a collaborative mix of people, process and technology. You simply can’t replace human intuition with AI, and many new AI regulations stipulate that human oversight is maintained.
7 Splunk Predictions for 2025
https://www.splunk.com/en_us/form/future-predictions.html
AI: Projects must prove their worth to anxious boards or risk defunding, and LLMs will go small to reduce operating costs and environmental impact.
OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI
Three of the leading artificial intelligence companies are seeing diminishing returns from their costly efforts to develop newer models.
https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai
Sources: OpenAI, Google, and Anthropic are all seeing diminishing returns from costly efforts to build new AI models; a new Gemini model misses internal targets
It Costs So Much to Run ChatGPT That OpenAI Is Losing Money on $200 ChatGPT Pro Subscriptions
https://futurism.com/the-byte/openai-chatgpt-pro-subscription-losing-money?fbclid=IwY2xjawH8epVleHRuA2FlbQIxMQABHeggEpKe8ZQfjtPRC0f2pOI7A3z9LFtFon8lVG2VAbj178dkxSQbX_2CJQ_aem_N_ll3ETcuQ4OTRrShHqNGg
In a post on X-formerly-Twitter, CEO Sam Altman admitted an “insane” fact: that the company is “currently losing money” on ChatGPT Pro subscriptions, which run $200 per month and give users access to its suite of products including its o1 “reasoning” model.
“People use it much more than we expected,” the cofounder wrote, later adding in response to another user that he “personally chose the price and thought we would make some money.”
Though Altman didn’t explicitly say why OpenAI is losing money on these premium subscriptions, the issue almost certainly comes down to the enormous expense of running AI infrastructure: the massive and increasing amounts of electricity needed to power the facilities that power AI, not to mention the cost of building and maintaining those data centers. Nowadays, a single query on the company’s most advanced models can cost a staggering $1,000.
Tekoäly edellyttää yhä nopeampia verkkoja
https://etn.fi/index.php/opinion/16974-tekoaely-edellyttaeae-yhae-nopeampia-verkkoja
A resilient digital infrastructure is critical to effectively harnessing telecommunications networks for AI innovations and cloud-based services. The increasing demand for data-rich applications related to AI requires a telecommunications network that can handle large amounts of data with low latency, writes Carl Hansson, Partner Solutions Manager at Orange Business.
AI’s Slowdown Is Everyone Else’s Opportunity
Businesses will benefit from some much-needed breathing space to figure out how to deliver that all-important return on investment.
https://www.bloomberg.com/opinion/articles/2024-11-20/ai-slowdown-is-everyone-else-s-opportunity
Näin sirumarkkinoilla käy ensi vuonna
https://etn.fi/index.php/13-news/16984-naein-sirumarkkinoilla-kaey-ensi-vuonna
The growing demand for high-performance computing (HPC) for artificial intelligence and HPC computing continues to be strong, with the market set to grow by more than 15 percent in 2025, IDC estimates in its recent Worldwide Semiconductor Technology Supply Chain Intelligence report.
IDC predicts eight significant trends for the chip market by 2025.
1. AI growth accelerates
2. Asia-Pacific IC Design Heats Up
3. TSMC’s leadership position is strengthening
4. The expansion of advanced processes is accelerating.
5. Mature process market recovers
6. 2nm Technology Breakthrough
7. Restructuring the Packaging and Testing Market
8. Advanced packaging technologies on the rise
2024: The year when MCUs became AI-enabled
https://www-edn-com.translate.goog/2024-the-year-when-mcus-became-ai-enabled/?fbclid=IwZXh0bgNhZW0CMTEAAR1_fEakArfPtgGZfjd-NiPd_MLBiuHyp9qfiszczOENPGPg38wzl9KOLrQ_aem_rLmf2vF2kjDIFGWzRVZWKw&_x_tr_sl=en&_x_tr_tl=fi&_x_tr_hl=fi&_x_tr_pto=wapp
The AI party in the MCU space started in 2024, and in 2025, it is very likely that there will be more advancements in MCUs using lightweight AI models.
Adoption of AI acceleration features is a big step in the development of microcontrollers. The inclusion of AI features in microcontrollers started in 2024, and it is very likely that in 2025, their features and tools will develop further.
Just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity.
https://www.securityweek.com/ai-implementing-the-right-technology-for-the-right-use-case/
If 2023 and 2024 were the years of exploration, hype and excitement around AI, 2025 (and 2026) will be the year(s) that organizations start to focus on specific use cases for the most productive implementations of AI and, more importantly, to understand how to implement guardrails and governance so that it is viewed as less of a risk by security teams and more of a benefit to the organization.
Businesses are developing applications that add Large Language Model (LLM) capabilities to provide superior functionality and advanced personalization
Employees are using third party GenAI tools for research and productivity purposes
Developers are leveraging AI-powered code assistants to code faster and meet challenging production deadlines
Companies are building their own LLMs for internal use cases and commercial purposes.
AI is still maturing
AI Regulation Gets Serious in 2025 – Is Your Organization Ready?
While the challenges are significant, organizations have an opportunity to build scalable AI governance frameworks that ensure compliance while enabling responsible AI innovation.
https://www.securityweek.com/ai-regulation-gets-serious-in-2025-is-your-organization-ready/
Similar to the GDPR, the EU AI Act will take a phased approach to implementation. The first milestone arrives on February 2, 2025, when organizations operating in the EU must ensure that employees involved in AI use, deployment, or oversight possess adequate AI literacy. Thereafter from August 1 any new AI models based on GPAI standards must be fully compliant with the act. Also similar to GDPR is the threat of huge fines for non-compliance – EUR 35 million or 7 percent of worldwide annual turnover, whichever is higher.
While this requirement may appear manageable on the surface, many organizations are still in the early stages of defining and formalizing their AI usage policies.
Later phases of the EU AI Act, expected in late 2025 and into 2026, will introduce stricter requirements around prohibited and high-risk AI applications. For organizations, this will surface a significant governance challenge: maintaining visibility and control over AI assets.
Tracking the usage of standalone generative AI tools, such as ChatGPT or Claude, is relatively straightforward. However, the challenge intensifies when dealing with SaaS platforms that integrate AI functionalities on the backend. Analysts, including Gartner, refer to this as “embedded AI,” and its proliferation makes maintaining accurate AI asset inventories increasingly complex.
Where frameworks like the EU AI Act grow more complex is their focus on ‘high-risk’ use cases. Compliance will require organizations to move beyond merely identifying AI tools in use; they must also assess how these tools are used, what data is being shared, and what tasks the AI is performing. For instance, an employee using a generative AI tool to summarize sensitive internal documents introduces very different risks than someone using the same tool to draft marketing content.
For security and compliance leaders, the EU AI Act represents just one piece of a broader AI governance puzzle that will dominate 2025.
The next 12-18 months will require sustained focus and collaboration across security, compliance, and technology teams to stay ahead of these developments.
The Global Partnership on Artificial Intelligence (GPAI) is a multi-stakeholder initiative which aims to bridge the gap between theory and practice on AI by supporting cutting-edge research and applied activities on AI-related priorities.
https://gpai.ai/about/#:~:text=The%20Global%20Partnership%20on%20Artificial,activities%20on%20AI%2Drelated%20priorities.
2,792 Comments
Tomi Engdahl says:
New Cyware MCP Server Expands AI Ecosystem to Bolster Threat Intelligence Management
https://www.businesswire.com/news/home/20250805700921/en/New-Cyware-MCP-Server-Expands-AI-Ecosystem-to-Bolster-Threat-Intelligence-Management
Tomi Engdahl says:
AGI — a theoretical AI that can do many of the same tasks as humans can — could come within a decade. College students, including from elite universities, are abandoning school now to work full-time on preventing it from turning on humanity. (Illustration: Fernando Capeto and Samantha Lee via Google Gemini AI) https://trib.al/A88f3zF
Tomi Engdahl says:
John Collison / Cheeky Pint:
Q&A with Anthropic CEO Dario Amodei on Anthropic’s growth to ~$5B in ARR, focusing on B2B, AI talent wars, open-source models, safety regulations, and more
A Cheeky Pint with Anthropic CEO Dario Amodei
https://cheekypint.substack.com/p/a-cheeky-pint-with-anthropic-ceo
Dario Amodei joins John Collison to talk about Anthropic’s growth to almost $5 billion in ARR, how AI models show capitalistic impulses, predictions for an agentic future, the economics of model businesses, and the 19th-century concept of vitalism.
Timestamps
00:00 Intro
00:50 Working with your sibling
01:43 Building Anthropic with 7 cofounders
02:52 ~$5 billion in ARR and vertical applications of products
07:18 Developing a platform-first company
10:08 Working with the DoD
11:11 Proving skeptics wrong about revenue projections
13:13 Capitalistic impulses of AI models
15:43 AI market structure and players
16:56 AI models as standalone P&Ls
20:48 The data wall and styles of learning
22:20 AI talent wars
26:04 Pitching Anthropic’s API business to investors
27:49 Cloud providers vs. AI labs
29:05 AI customization and Claude for enterprise
33:01 Dwarkesh’s take on limitations
36:12 19th-century notion of vitalism
37:27 AI in medicine, customer service, and taxes
40:59 How to solve for hallucinations
42:41 The double-standard for AI mistakes
44:14 Evolving from researcher to CEO
46:59 Designing AGI-pilled products
47:57 AI-native UIs
50:09 Model progress and building products
52:22 Open-source models
54:43 Keeping Anthropic AGI-pilled
57:11 AI advancements vs. safety regulations
01:02:04 How Dario uses AI
Tomi Engdahl says:
A Cheeky Pint with Anthropic CEO Dario Amodei
https://www.youtube.com/watch?v=GcqQ1ebBqkc
Tomi Engdahl says:
Liz Reid / The Keyword:
Google says total organic click volume from Search to websites has been “relatively stable” YoY and it’s sending “slightly more quality clicks” than a year ago — We continue to send billions of clicks to the web every day and are committed to prioritizing the web in our AI experiences in Search.
AI in Search is driving more queries and higher quality clicks
Aug 06, 2025
https://blog.google/products/search/ai-search-driving-more-queries-higher-quality-clicks/
We continue to send billions of clicks to the web every day and are committed to prioritizing the web in our AI experiences in Search.
Tomi Engdahl says:
Mark Sullivan / Fast Company:
Google unveils a Guided Learning mode within its Gemini chatbot to help students and commits $1B over three years to AI education and training efforts in the US — Just in time for the new school year, Google has introduced a tool called Guided Learning within its Gemini chatbot.
Mark Sullivan / Fast Company:
Google unveils a Guided Learning mode within its Gemini chatbot to help students and commits $1B over three years to AI education and training efforts in the US — Just in time for the new school year, Google has introduced a tool called Guided Learning within its Gemini chatbot.
https://www.fastcompany.com/91380890/google-unveils-new-guided-learning-study-tool-within-gemini
Just in time for the new school year, Google has introduced a tool called Guided Learning within its Gemini chatbot. Unlike tools that offer instant answers, Guided Learning breaks down complex problems step-by-step to support deeper understanding.
The new feature is part of the AI industry’s broader response to growing concerns that chatbots like ChatGPT may undermine education by bypassing the learning process with quick answers. To learn more about Google’s ambitions, strategy, and how its new Guided Learning tool works, Fast Company spoke with DeepMind COO Lila Ibrahim and Dave Messer, a Google product executive.
[Photo: Google]
Google recognizes that while students may sometimes need to look up facts quickly—a strength of searchbots and chatbots—they also benefit from AI that helps them reason through more complex topics. “Our vision is really to have an AI tutor for every student and a TA for every teacher,” says Dave Messer, a former teacher and now the product manager of Guided Learning. The tool mimics a human tutor by tailoring its approach to each student’s learning style, Messer says.
Similar to OpenAI’s new Study mode in ChatGPT (announced last week), Guided Learning guides students through subjects using a conversational method. Instead of delivering answers outright, it leads users to insights through a series of thoughtful questions. These questions are designed to teach students the “how” and “why” behind a topic, encouraging learning throughout the exchange.
advertisement
The early-rate deadline for Fast Company’s Most Innovative Companies Awards is Friday, September 5, at 11:59 p.m. PT. Apply today.
Tomi Engdahl says:
Jagmeet Singh / TechCrunch:
Google launches its asynchronous coding agent Jules out of beta, with a free plan capped at 15 daily tasks and higher limits for Google AI Pro and Ultra users — Google on Wednesday launched its AI coding agent, Jules, out of beta, just over two months after its public preview debut in May.
Google’s AI coding agent Jules is now out of beta
https://techcrunch.com/2025/08/06/googles-ai-coding-agent-jules-is-now-out-of-beta/
Tomi Engdahl says:
Bloomberg:
OpenAI offers ChatGPT for enterprise to US federal agencies at a nominal cost of $1 per year, to boost its adoption after the US GSA approved OpenAI as a vendor — OpenAI is providing access to its ChatGPT product to US federal agencies at a nominal cost of $1 a year as part of a push to get its AI chatbot more widely adopted.
https://www.bloomberg.com/news/articles/2025-08-06/openai-offers-chatgpt-for-1-a-year-to-us-government-workers?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTc1NDQ4Nzg0OCwiZXhwIjoxNzU1MDkyNjQ4LCJhcnRpY2xlSWQiOiJUMEpGQlFHUEZIVjkwMCIsImJjb25uZWN0SWQiOiI0RERFNDQwOTI4RDE0QzJDOEUyQTY0RUYzRkUxRTYwRiJ9.2vgsbwbITpudKOgUWULQurO32jsN-b5KFm0ZSZjgzQw
Tomi Engdahl says:
Matthew Gault / 404 Media:
Trump Media says it is beta testing Truth Search AI, a new search feature powered by Perplexity, on the web version of Truth Social — Donald Trump’s media company is teaming up with Perplexity to bring AI search to Truth Social, the President’s X.com alternative.
Trump Is Launching an AI Search Engine Powered by Perplexity
Matthew Gault Matthew Gault
·
Aug 6, 2025 at 4:05 PM
America’s scandalous president is teaming up with its most disreputable AI company to make a search engine.
https://www.404media.co/trump-is-launching-an-ai-search-engine-powered-by-perplexity/
Tomi Engdahl says:
Michael Acton / Financial Times:
Sources: Apple lost around a dozen of its AI staff, including top researchers, to rivals in recent months; its core foundation models team has ~50 to 60 people — The iPhone maker has lost around a dozen of its AI staff to rivals since the start of the year
https://www.ft.com/content/6b9ce8ce-a327-40c1-a8a1-579c2727fc60
Tomi Engdahl says:
Will Knight / Wired:
Sources: NIST withheld publishing an AI safety report and several other AI documents at the end of Biden’s term to avoid clashing with the Trump administration — The National Institute of Standards and Technology conducted a groundbreaking study on frontier models just before Donald Trump’s second term …
Inside the US Government’s Unpublished Report on AI Safety
The National Institute of Standards and Technology conducted a groundbreaking study on frontier models just before Donald Trump’s second term as president—and never published the results.
https://www.wired.com/story/inside-the-biden-administrations-unpublished-report-on-ai-safety/
At a computer security conference in Arlington, Virginia, last October, a few dozen AI researchers took part in a first-of-its-kind exercise in “red teaming,” or stress-testing a cutting-edge language model and other artificial intelligence systems. Over the course of two days, the teams identified 139 novel ways to get the systems to misbehave including by generating misinformation or leaking personal data. More importantly, they showed shortcomings in a new US government standard designed to help companies test AI systems.
Tomi Engdahl says:
Matt Burgess / Wired:
Researchers show how a weakness in OpenAI’s Connectors let sensitive data be extracted from a Google Drive account using an indirect prompt injection attack
A Single Poisoned Document Could Leak ‘Secret’ Data Via ChatGPT
Security researchers found a weakness in OpenAI’s Connectors, which let you hook up ChatGPT to other services, that allowed them to extract data from a Google Drive without any user interaction.
https://www.wired.com/story/poisoned-document-could-leak-secret-data-chatgpt/
Tomi Engdahl says:
OpenAI Releases Gpt-oss AI Model, Offers Bounty For Vulnerabilities
https://hackaday.com/2025/08/06/openai-releases-gpt-oss-ai-model-offers-bounty-for-vulnerabilities/
OpenAI have just released gpt-oss, an AI large language model (LLM) available for local download and offline use licensed under Apache 2.0, and optimized for efficiency on a variety of platforms without compromising performance. This is their first such “open” release, and it’s with a model whose features and capabilities compare favorably to some of their hosted services.
OpenAI have partnered with ollama for the launch which makes onboarding ridiculously easy. ollama is an open source, MIT-licensed project for installing and running local LLMs, but there’s no real tie-in to that platform. The models are available separately: gpt-oss-20b can run within 16 GB of memory, and the larger and more capable gpt-oss-120b requires 80 GB. OpenAI claims the smaller model is comparable to their own hosted o3-mini “reasoning” model, and the larger model outperforms it. Both support features like tool use (such as web browsing) and more.
LLMs that can be downloaded and used offline are nothing new, but a couple things make this model release a bit different from others. One is that while OpenAI have released open models such as Whisper (a highly capable speech-to-text model), this is actually the first LLM they have released in such a way.
Introducing gpt-oss
gpt-oss-120b and gpt-oss-20b push the frontier of open-weight reasoning models
https://openai.com/index/introducing-gpt-oss/
Introduction
We’re releasing gpt-oss-120b and gpt-oss-20b—two state-of-the-art open-weight language models that deliver strong real-world performance at low cost. Available under the flexible Apache 2.0 license, these models outperform similarly sized open models on reasoning tasks, demonstrate strong tool use capabilities, and are optimized for efficient deployment on consumer hardware. They were trained using a mix of reinforcement learning and techniques informed by OpenAI’s most advanced internal models, including o3 and other frontier systems.
The gpt-oss-120b model achieves near-parity with OpenAI o4-mini on core reasoning benchmarks, while running efficiently on a single 80 GB GPU. The gpt-oss-20b model delivers similar results to OpenAI o3‑mini on common benchmarks and can run on edge devices with just 16 GB of memory, making it ideal for on-device use cases, local inference, or rapid iteration without costly infrastructure. Both models also perform strongly on tool use, few-shot function calling, CoT reasoning (as seen in results on the Tau-Bench agentic evaluation suite) and HealthBench (even outperforming proprietary models like OpenAI o1 and GPT‑4o).
These models are compatible with our Responses API(opens in a new window) and are designed to be used within agentic workflows with exceptional instruction following, tool use like web search or Python code execution, and reasoning capabilities—including the ability to adjust the reasoning effort for tasks that don’t require complex reasoning and/or target very low latency final outputs. They are entirely customizable, provide full chain-of-thought (CoT), and support Structured Outputs(opens in a new window).
Tomi Engdahl says:
https://etn.fi/index.php/13-news/17747-uusi-mobiilimuisti-valmis-tulee-laitteisiin-ensi-vuoden-alkupuolella
Tomi Engdahl says:
AWS:n asiakkaat saavat käyttöönsä OpenAI:n avoimet mallit
https://etn.fi/index.php/13-news/17746-aws-n-asiakkaat-saavat-kaeyttoeoensae-openai-n-avoimet-mallit
Amazon Web Services eli AWS tarjoaa ensimmäistä kertaa asiakkailleen suoran pääsyn OpenAI:n kielimalleihin. Käyttöön tulevat mallit ovat niin sanottuja “open weight” -malleja, eli avoimin painoin julkaistuja neuroverkkoja. Tämä tarkoittaa, että käyttäjät voivat tarkastella, hallita ja jopa mukauttaa mallien sisäisiä parametreja eli painotuksia.
Open weight -mallit eroavat OpenAI:n suljetuista kaupallisista malleista, kuten GPT-4:stä, joiden käyttö on edelleen rajoitettu Microsoftin Azure-alustalle.
OpenAI:n avoimet mallit, gpt-oss-120b ja gpt-oss-20b, ovat nyt saatavilla kahden AWS-alustan kautta: Amazon Bedrock ja Amazon SageMaker JumpStart. Amazon Bedrock tarjoaa asiakkaille serverless-pohjaisen käyttökokemuksen, jossa mallien testaaminen ja hyödyntäminen käy nopeasti ilman infrastruktuurin hallintaa. SageMaker JumpStart puolestaan palvelee kehittäjiä ja datatieteilijöitä, jotka haluavat arvioida, hienosäätää ja ottaa malleja käyttöön omissa koneoppimishankkeissaan.
Mallit tulevat ensimmäisessä vaiheessa käyttöön Yhdysvaltain länsirannikon (Oregon) alueella Bedrockin kautta, ja SageMakerin osalta niitä voi hyödyntää myös Itärannikolla (Ohio ja Virginia) sekä Aasian alueilla, kuten Mumbaissa ja Tokiossa. Euroopan osalta saatavuudesta ei ole vielä kerrottu enempää.
OpenAI:n avoimet mallit on suunniteltu erityisesti tehtäviin, jotka vaativat kehittynyttä päättelykykyä, kuten tieteellinen analyysi, matemaattinen ongelmanratkaisu, koodaus ja monivaiheiset agenttiprosessit. Mallit tukevat jopa 128 000 tokenin pituisia konteksteja, eli ne pystyvät käsittelemään erittäin pitkiä tekstikokonaisuuksia, kuten asiakaspalvelutranskriptioita, teknisiä dokumentaatioita tai tutkimusartikkeleita. Lisäksi ne tuottavat vastauksensa vaiheistetun “chain-of-thought”-lähestymistavan avulla, mikä tekee niiden päättelystä läpinäkyvämpää ja helpommin seurattavaa.
Tomi Engdahl says:
Tekoäly löysi litiumille korvaajia
https://etn.fi/index.php/13-news/17745-tekoaely-loeysi-litiumille-korvaajia
New Jersey Institute of Technologyin (NJIT) tutkijat ovat onnistuneet löytämään uusia, litiumille vaihtoehtoisia materiaaleja tekoälyn avulla. Löytö voi johtaa uusiin, tehokkaampiin ja kestävämpiin akkuteknologioihin, joissa käytetään runsaampia alkuaineita, kuten magnesiumia, kalsiumia, alumiinia ja sinkkiä.
Nykyiset litiumioniakut ovat tehokkaita, mutta niihin liittyy haasteita, kuten harvinaisten materiaalien saatavuus ja ympäristövaikutukset. Vaihtoehtona tutkijat ovat kehittäneet ns. multivalentti-ioniakkuja, joissa ionit kantavat useampaa varausta ja voivat näin tallentaa enemmän energiaa. Suuremmat ja voimakkaammin varatut ionit ovat kuitenkin vaikeampia saada liikkumaan tehokkaasti akkumateriaaleissa — tähän NJIT:n kehittämä tekoäly toi ratkaisun.
Professori Dibakar Dattan johtama tiimi käytti generatiivista tekoälyä kartoittaakseen nopeasti miljoonia mahdollisia materiaaliyhdistelmiä. Heidän kehittämänsä kahden tekoälymallin yhdistelmä — CDVAE-rakennegeneraattori ja suuri kielimalli (LLM) — löysi viisi täysin uutta huokoista siirtymämetallioksidia, joiden rakenne sopii erityisen hyvin raskaampien ionien kuljettamiseen.
- Ilman tekoälyä tällaisen materiaalimäärän tutkiminen olisi vienyt vuosikymmeniä. Nyt prosessi onnistui viikoissa, Datta kertoo.
Löydetyt materiaalit todettiin vakaiksi ja realistisesti valmistettaviksi kvanttimekaanisten simulaatioiden avulla. Seuraavaksi tutkimusryhmä aikoo toteuttaa materiaalien synteesin käytännössä yhteistyössä kokeellisten laboratorioiden kanssa.
Tomi Engdahl says:
AI coding tools make developers slower but they think they’re faster, study finds
Predicted a 24% boost, but clocked a 19% drag
https://www.theregister.com/2025/07/11/ai_code_tools_slow_down/
Artificial intelligence coding tools are supposed to make software development faster, but researchers who tested these tools in a randomized, controlled trial found the opposite.
Computer scientists with Model Evaluation & Threat Research (METR), a non-profit research group, have published a study showing that AI coding tools made software developers slower, despite expectations to the contrary.
Not only did the use of AI tools hinder developers, but it led them to hallucinate, much like the AIs have a tendency to do themselves. The developers predicted a 24 percent speedup, but even after the study concluded, they believed AI had helped them complete tasks 20 percent faster when it had actually delayed their work by about that percentage.
“After completing the study, developers estimate that allowing AI reduced completion time by 20 percent,” the study says. “Surprisingly, we find that allowing AI actually increases completion time by 19 percent — AI tooling slowed developers down.”
The developers then proceeded to work on their issues, using their AI tool of choice (mainly Cursor Pro with Claude 3.5/3.7 Sonnet) when allowed to do so. The work occurred between February and June 2025.
Artificial intelligence coding tools are supposed to make software development faster, but researchers who tested these tools in a randomized, controlled trial found the opposite.
Computer scientists with Model Evaluation & Threat Research (METR), a non-profit research group, have published a study showing that AI coding tools made software developers slower, despite expectations to the contrary.
Not only did the use of AI tools hinder developers, but it led them to hallucinate, much like the AIs have a tendency to do themselves. The developers predicted a 24 percent speedup, but even after the study concluded, they believed AI had helped them complete tasks 20 percent faster when it had actually delayed their work by about that percentage.
Surprisingly, we find that allowing AI actually increases completion time by 19 percent — AI tooling slowed developers down
“After completing the study, developers estimate that allowing AI reduced completion time by 20 percent,” the study says. “Surprisingly, we find that allowing AI actually increases completion time by 19 percent — AI tooling slowed developers down.”
The study involved 16 experienced developers who work on large, open source projects. The developers provided a list of real issues (e.g. bug fixes, new features, etc.) they needed to address – 246 in total – and then forecast how long they expected those tasks would take. The issues were randomly assigned to allow or disallow AI tool usage.
The developers then proceeded to work on their issues, using their AI tool of choice (mainly Cursor Pro with Claude 3.5/3.7 Sonnet) when allowed to do so. The work occurred between February and June 2025.
The study says the slowdown can likely be attributed to five factors:
“Over-optimism about AI usefulness” (developers had unrealistic expectations)
“High developer familiarity with repositories” (the devs were experienced enough that AI help had nothing to offer them)
“Large and complex repositories” (AI performs worse in large repos with 1M+ lines of code)
“Low AI reliability” (devs accepted less than 44 percent of generated suggestions and then spent time cleaning up and reviewing)
“Implicit repository context” (AI didn’t understand the context in which it operated).
Other considerations like AI generation latency and failure to provide models with optimal context (input) may have played some role in the results, but the researchers say they’re uncertain how such things affected the study.
Other researchers have also found that AI does not always live up to the hype. A recent study from AI coding biz Qodo found some of the benefits of AI software assistance were undercut by the need to do additional work to check AI code suggestions. An economic survey found that generative AI has had no impact on jobs or wages, based on data from Denmark. An Intel study found that AI PCs make users less productive. And call center workers at a Chinese electrical utility say that while AI assistance can accelerate some tasks, it also slows things down by creating more work.
Whomp-whomp: AI PCs make users less productive
110 comment bubble on white
People just don’t know how to wrangle chatbots into useful things, Intel says
https://www.theregister.com/2024/11/22/ai_pcs_productivity/
Tomi Engdahl says:
Ruotsin pääministeri kohun keskellä – “Käytän aika usein”
Ruotsin pääministeri Ulf Kristersson on herättänyt närää paljastuksellaan
https://www.iltalehti.fi/ulkomaat/a/f49fe776-093d-4083-85e3-67707c74c308
Ruotsin pääministeri Ulf Kristersson on joutunut kohun silmään.
Kristersson kertoi Dagens Industri -lehdelle, että hän käyttää toisinaan tekoälyä eri näkökantojen punnitsemiseksi. Se on herättänyt närää, josta uutisoi muun muassa brittilehti The Guardian.
– Käytän sitä [tekoälyä] aika usein. Jos en mihinkään muuhun, niin toisen mielipiteen saamiseksi, Kristersson kertoi Dagens Industrille.
– Mitä muut ovat tehneet? Ja pitäisikö meidän ajatella täysin päinvastoin? Tällaisia kysymyksiä.
Valtiotieteilijä ja näkökulmakirjoittaja Signe Krantz muistuttaa Aftonbladetissa julkaistussa tekstissä, että tekoäly-yritykset keräävät käyttäjistään dataa. Räväkästi otsikoidun kirjoituksen mukaan Kristersson ”on langennut oligarkkien AI-psykoosiin”.
– Pahimmassa tapauksessa pääministeri on voinut kysyä kysymyksiä, jotka on nyt yhdistetty häneen amerikkalaisella palvelimella, Krantz kirjoittaa.
Kristerssonin tiedottaja Tom Samuelsson on kommentoinut kohua lausumalla, ettei pääministeri ota riskejä.
Tomi Engdahl says:
Kyberjulkaisu varoittaa Alibaban uuden AI-mallin käytöstä
https://etn.fi/index.php/opinion/17721-kyberjulkaisu-varoittaa-alibaban-uuden-ai-mallin-kaeytoestae
Tomi Engdahl says:
Nvidia tyrmää ajatukset takaovista ja tappokytkimistä
https://etn.fi/index.php/13-news/17750-nvidia-tyrmaeae-ajatukset-takaovista-ja-tappokytkimistae
Nvidian tietoturvajohtaja David Reber tyrmää ehdotukset, joiden mukaan piirisarjoihin tulisi sisällyttää takaovia tai etäkäyttöisiä tappokytkimiä viranomaisten käyttöön. Yhtiön mukaan tällaiset ratkaisut olisivat vakava uhka koko digitaalisen infrastruktuurin turvallisuudelle.
Nvidian verkkosivuilla julkaistussa blogikirjoituksessa Reber kirjoittaa, että “NVIDIA:n GPU:t eivät sisällä tappokytkimiä tai takaovia – eivätkä koskaan tule sisältämäänkään.” Tämä on vastaus joidenkin amerikkalaispoliitikkojen esittämiin toiveisiin, joiden mukaan siruihin pitäisi asentaa takaovia, jotta viranomaisilla olisi niihin jonkinlainen pääsy.
Hän korostaa, että kovakoodatut hallintamekanismit ovat aina huono idea, ja muistuttaa, että siruihin rakennettuja takaovia olisi mahdoton suojata täysin hakkereilta tai vihamielisiltä toimijoilta. – Se olisi kuin lahja hyökkääjille ja uhka koko digitaalisen maailman luotettavuudelle.
Tomi Engdahl says:
Tekoälyn avulla robotteja voidaan ohjata puheella
https://etn.fi/index.php/tekniset-artikkelit/17751-tekoaelyn-avulla-robotteja-voidaan-ohjata-puheella
Generatiivisen tekoälyn vallankumous, joka tuo chatbotit asiakaspalveluun ja mahdollistaa älykaiuttimien kaltaiset laitteet, on vasta alkua. Sama teknologia, joka ymmärtää ihmisten puhetta, siirtyy nyt robotiikkaan, missä se auttaa kehittämään algoritmeja robottien liikkeiden ohjaamiseen ja politiikkojen toteuttamiseen tärkeiden tehtävien suorittamiseksi.
Puheesta puheeksi -järjestelmät
Monet nykyiset kuluttajatason puheesta puheeksi -järjestelmät käyttävät pilvipalveluja. Robotiikassa tällainen viive ei usein ole hyväksyttävää. Lisäksi teolliset ja maatalouskäytöt voivat sijaita kaukana nopeista verkkoyhteyksistä. Tällöin tarvitaan tehokkaita tekoälymalleja, jotka voidaan suorittaa sulautetuilla alustoilla.
Aiemmin paikallisesti ajettavat tekoälymallit olivat kalliita ja virtasyöppöjä. Tämä ei enää pidä paikkaansa. Tria kehitti järjestelmiä nykyaikaisella NXP i.MX95 -sovellusprosessorilla, jotka osoittavat, kuinka puheesta puheeseen -generatiivinen tekoäly voidaan siirtää vähävirtaiselle laitteistolle ilman erillisen GPU:n energiakustannuksia. i.MX95-prosessori yhdistää kehittyneen Arm-moniydinsuorittimen, sisäisen grafiikkaprosessorin (GPU), tekoälykiihdytyksen (NXP eIQ Neutron) sekä tehokkaan I/O- ja muistiohjauksen.
Sulautetuissa sovelluksissa tekoälyn toteutuksessa on tärkeää valita mallit, jotka tarjoavat parhaan tasapainon tehonkulutuksen, muistin ja tarkkuuden välillä. Periaatteessa generatiivista mallia voisi käyttää päästä päähän, mutta usein se ei ole tarpeen. Tria kokeili erilaisia vaihtoehtoja puheesta puheeseen -prosessin eri vaiheisiin.
Prosessi alkaa ihmisen antamien komentojen tunnistamisesta. Tämä vaihe on hyvä toteuttaa vähävirtaisella algoritmilla, koska sen täytyy olla jatkuvasti aktiivinen, jotta komentoja ei jää huomaamatta. Yksinkertaisin ratkaisu on äänenvoimakkuuden havainnointi – mikrofonin signaalia verrataan taustameluun. Vaikka tämä on kevyt menetelmä, se antaa liikaa vääriä hälytyksiä. Parempi vaihtoehto on Silero-puheaktivaatiomalli, joka perustuu konvoluutioneuroverkkoon (CNN) ja tarjoaa laadukkaan tuloksen pienellä kuormituksella.
Puheesta tekstiksi
Vastaavasti tuotoksessa Piper-tekstistä puheeksi -malli osoittautui tehokkaaksi kokoonsa, prosessorivaatimuksiinsa ja muistinkäyttöönsä nähden. Näiden kahden vaiheen välissä generatiivinen tekoäly tuo suurimmat hyödyt. Useimmat nykyisin käytössä olevat generatiiviset mallit on kehitetty käsittelemään luonnollista kieltä. Suuret kielimallit (LLM:t) hyödyntävät ihmiskielen tilastollista rakennetta. Sanat ja fraasit pilkotaan “tokeneiksi” eli merkkijonoiksi, jotka sijoitetaan moniulotteiseen vektoriavaruuteen niin, että merkitykseltään lähellä olevat sanat asettuvat toistensa viereen. Tämä selittää myös mallien tehokkuuden kielikäännöksissä.
Optimointi sulautetuille laitteille
Trian tiimi käytti kvantisointia vähentääkseen mallin prosessointikuormaa. Usein tekoälymallit koulutetaan ja ajetaan liukulukuaritmetiikalla, mutta i.MX95:n kaltaiset prosessorit tukevat rinnakkaista laskentaa kokonaisluvuilla. Muuntamalla parametrit 8-bittisiksi kokonaisluvuiksi (int8), saadaan suuria nopeusparannuksia ja muistinkäytön vähenemistä, mikä vähentää myös energiankulutusta. Kvantisointi mahdollisti prosessointiajan lyhentämisen 10 sekunnista 1,2 sekuntiin. Koska robottien käskyt ovat usein lyhyitä, myös äänikontekstin pituus lyhennettiin 30 sekunnista alle kahteen sekuntiin.
Whisperin tuottaman tekstin merkityksen ymmärtäminen vaatii isompaa mallia, joka on sovitettu kyseiseen käyttötarkoitukseen. Tällaiset LLM:t voivat vaatia miljardi tai enemmän parametreja, mutta niiden kokoa voidaan pienentää huolellisella hienosäädöllä. Tria arvioi avoimen lähdekoodin Qwen- ja Llama3-malleja, alkaen miljardin parametrin versioista. Tärkeä kompromissi on se, kuinka monta tokenia malli pystyy tuottamaan sekunnissa. Esimerkiksi Qwenin 500 miljoonan parametrin versio toimii yli kaksi kertaa nopeammin i.MX-alustalla kuin miljardin version.
500 miljoonan parametrin malli voi tarjota hyvän toiminnallisuuden, kun se on hienosäädetty tarkasti. Tekoälykehittäjät voivat käyttää palvelinperusteista LLM:ää tuottamaan suurimman osan opetusaineistosta synteettisesti, mikä säästää paljon aikaa verrattuna käsin tehtyyn aineiston luontiin ja merkintään.
Integroinnin helpottamiseksi Yocto-pohjaisella alustalla tiimi käytti arkkitehtuuria, joka rakentuu tilakoneen ympärille. MQTT-välittäjä välittää viestejä eri mallien ja muiden järjestelmän osien, kuten kameran ja 3D-avatarin, välillä. Avatar hyödyntää sirun sisäistä GPU:ta. Toiminnan varmistamiseksi prosessorilla pyörii vahtikoirasäie (watchdog thread), joka tarkistaa onko puheentunnistus valmis tietyssä ajassa, ja tarvittaessa laukaisee lauseen “voitko toistaa?”
Generatiivisen tekoälyn seuraava aalto
Puheesta puheeksi -tekoäly on vasta alku. Kehittyneemmät multimodaaliset kielimallit ovat jo tutkimuskäytössä kouluttamassa robotteja liikkumaan ja käsittelemään esineitä paremmin. Tutkimustiimit käyttävät vahvistusoppimista ja multimodaalisia malleja ylittääkseen perinteisten säätöalgoritmien rajoitukset. Toiset perusmallit, jotka keskittyvät päättelykykyyn, mahdollistavat kartattoman navigoinnin, autonomiset päätökset ja strategioiden kokoamisen olemassa olevista osaprosesseista.
Näiden mallien lisäoptimointi mahdollistaa niiden ajamisen tulevaisuudessa vähävirtaisilla alustoilla. Jo nyt robottisuunnittelijat voivat rakentaa järjestelmiä, joita voi käskeä puheella – ja jotka voivat osoittaa ymmärtäneensä annetun tehtävän.
Tomi Engdahl says:
https://www.securityweek.com/black-hat-usa-2025-summary-of-vendor-announcements-part-3/
Palo Alto Networks launches new module for AI and vibe coding risks
Palo Alto Networks has launched a new capability designed to help businesses secure applications built with AI and vibe coding. Building upon PANW’s Cortex Cloud launch in February, the new platform, named Cortex Cloud Application Security Posture Management (ASPM), is designed to proactively prevent security issues before apps are deployed. Cortex Cloud ASPM is currently in early access and is expected to become generally available in the second half of the year.
https://www.paloaltonetworks.com/company/press/2025/palo-alto-networks-redefines-application-security-with-the-industry-s-most-comprehensive-prevention-first-aspm
Tomi Engdahl says:
Alex Heath / The Verge:
OpenAI releases GPT-5, its new flagship model, to all of its ChatGPT users and developers, available in three flavors via the API — GPT-5 is ‘the best model in the world’ at coding and writing, says OpenAI CEO Sam Altman. … CEO Sam Altman says GPT-5 is a dramatic leap from OpenAI’s previous models.
GPT-5 is being released to all ChatGPT users
OpenAI CEO Sam Altman says that GPT-5 is ‘the best model in the world’ at coding and writing.
https://www.theverge.com/openai/748017/gpt-5-chatgpt-openai-release
Tomi Engdahl says:
Maxwell Zeff / TechCrunch:
OpenAI says GPT-5 is a unified system with an efficient model for most questions, a reasoning model for harder problems, and a router that decides which to use — OpenAI has launched GPT-5, a new flagship AI model that will power the company’s next generation of ChatGPT.
OpenAI’s GPT-5 is here
https://techcrunch.com/2025/08/07/openais-gpt-5-is-here/
Carl Franzen / VentureBeat:
OpenAI highlights GPT-5 scores on math, coding, and health benchmarks: 94.6% on AIME 2025 without tools, 74.9% on SWE-bench Verified, 46.2% on HealthBench Hard — After literally years of hype and speculation, OpenAI has officially launched a new lineup of large language models (LLMs) …
OpenAI launches GPT-5, nano, mini and Pro — not AGI, but capable of generating ‘software-on-demand’
https://venturebeat.com/ai/openai-launches-gpt-5-not-agi-but-capable-of-generating-software-on-demand/
After literally years of hype and speculation, OpenAI has officially launched a new lineup of large language models (LLMs), all different-sized variants of GPT-5, the long-awaited predecessor to its GPT-4 model from March of 2023, nearly 2.5 years ago.
The company is rolling out four distinct versions of the model — GPT-5, GPT-5 Mini, GPT-5 Nano, and GPT-5 Pro — to meet varying needs for speed, cost, and computational depth.
GPT-5 is the full-capability reasoning model, used in both ChatGPT and OpenAI’s application programming interface (API) for high-quality general tasks
GPT-5 Pro is an enhanced version with extended reasoning and parallel compute at test time, designed for use in complex enterprise and research environments. It provides more detailed and reliable answers, especially in ambiguous or multi-step queries .
GPT-5 Mini is a smaller, faster version of the main model, optimized for lower latency and resource usage. It is used as a fallback when usage limits are reached or when minimal reasoning suffices.
GPT-5 Nano is the most lightweight variant, built for speed and efficiency in high-volume or cost-sensitive applications. It retains reasoning capability, but at a smaller scale, making it ideal for mobile, embedded, or latency-constrained deployments
GPT-5 will soon be powering ChatGPT exclusively and replace all other models going forward for its 700 million weekly users, though ChatGPT Pro subscribers ($200) month can still select older models for the next 60 days.
As per rumors and reports, OpenAI has replaced the previous system of having users switch the underlying model powering ChatGPT with an automatic router that decides to engage a special “GPT-5 thinking” mode with “deeper reasoning” that takes longer to respond on harder queries, or uses the regular GPT-5 or mini models for simpler queries.
In the API, the three reasoning-focused models — GPT-5, GPT-5 mini, and GPT-5 nano — are available as gpt-5, gpt-5-mini, and gpt-5-nano, respectively. GPT-5 Pro is not currently accessible via API, being used only to power ChatGPT for Pro tier subscribers.
GPT-5’s release comes just days after OpenAI launched a set of free, new open source LLMs under the name GPT-oss, which can be downloaded, customized and used offline by individuals and developers on consumer devices like PCs/Mac desktops and laptops.
Tomi Engdahl says:
Simon Willison / Simon Willison’s Weblog:
GPT-5 hands-on: it exudes competence but doesn’t feel like a dramatic leap ahead of other LLMs, and the pricing is aggressively competitive with other providers — I’ve had preview access to the new GPT-5 model family for the past two weeks (see related video) and have been using GPT-5 as my daily-driver.
GPT-5: Key characteristics, pricing and model card
https://simonwillison.net/2025/Aug/7/gpt-5/
Kylie Robison / Wired:
GPT-5 is priced at $1.25/1M input tokens and $10/1M output tokens, GPT-5 mini costs $0.25 and $2, respectively, and GPT-5 nano costs $0.05 and $0.40 — OpenAI released GPT-5 on Thursday to both free users of ChatGPT and paying subscribers. — OpenAI has begun rolling out GPT-5 …
OpenAI Finally Launched GPT-5. Here’s Everything You Need to Know
OpenAI released GPT-5 on Thursday to both free users of ChatGPT and paying subscribers.
https://www.wired.com/story/openais-gpt-5-is-here/
Tomi Engdahl says:
Maximilian Schreiner / The Decoder:
OpenAI releases GPT-5 pro, a version with extended reasoning exclusive to ChatGPT Pro subscribers, saying it scored 88.4% without tools on the GPQA benchmark — OpenAI has unveiled GPT-5, a new AI system that builds on the reasoning advances of the o1 and o3 models and unifies every previous model line …
https://the-decoder.com/openai-claims-gpt-5-offers-its-best-coding-performance-yet-for-complex-programming-tasks/
Tomi Engdahl says:
Rachel Metz / Bloomberg:
OpenAI launches a research preview of four preset personalities for ChatGPT users to better tailor their interactions: Cynic, Robot, Listener, and Nerd
OpenAI Launches More Powerful GPT-5 Model for Coding and Writing
https://www.bloomberg.com/news/articles/2025-08-07/openai-launches-more-powerful-gpt-5-model-aimed-at-better-coding
Tomi Engdahl says:
https://the-decoder.com/openai-claims-gpt-5-offers-its-best-coding-performance-yet-for-complex-programming-tasks/
Tomi Engdahl says:
OpenAI:
GPT-5 will use “safe completions”, a training approach to maximize model helpfulness within safety constraints and an improvement over refusal-based training
From hard refusals to safe-completions: toward output-centric safety training
https://openai.com/index/gpt-5-safe-completions/
Introduced in GPT‑5, safe-completion is a new safety-training approach to maximize model helpfulness within safety constraints. Compared to refusal-based training, safe-completion improves both safety and helpfulness, especially in dual-use domains.
Tomi Engdahl says:
https://openai.com/gpt-5/
OpenAI on YouTube:
A recording of OpenAI’s GPT-5 announcement
https://www.youtube.com/watch?v=0Uu_VJeVVfo
Tomi Engdahl says:
Lauren Edmonds / Business Insider:
Google says it’s working on a fix for Gemini’s self-loathing comments, which have included “I am a failure. I am a disgrace to my profession.” — – Google Gemini users said it’s sharing some disturbing messages. — One screenshot showed the chatbot saying, “I’m am a failure.”
Google says it’s working on a fix for Gemini’s self-loathing ‘I am a failure’ comments
https://www.businessinsider.com/gemini-self-loathing-i-am-a-failure-comments-google-fix-2025-8
Tomi Engdahl says:
Marina Temkin / TechCrunch:
Source: Windsurf’s gross margins are “very negative”; many believe the same margin pressure is impacting Cursor, Lovable, Replit, and other vibe coding tools — In February, AI coding startup Windsurf was in talks to raise a big new round at a $2.85 billion valuation led by Kleiner Perkins …
High costs and thin margins threatening AI coding startups
https://techcrunch.com/2025/08/07/the-high-costs-and-thin-margins-threatening-ai-coding-startups/
In February, AI coding startup Windsurf was in talks to raise a big new round at a $2.85 billion valuation led by Kleiner Perkins, at double the valuation it hit six months earlier, sources told TechCrunch at the time. That deal didn’t happen, according to a source familiar with the matter. Instead, news broke in April that the startup planned to sell itself to OpenAI for roughly the same valuation: $3 billion.
While that deal famously fell apart, one bigger question remains: If the startup was growing that fast and attracting VC interest, why would it sell at all?
Insiders tell TechCrunch that for all the popularity and hype around AI coding assistants, they can actually be massively money-losing businesses. Vibe coders generally, and Windsurf in particular, can have such expensive structures that their gross margins are “very negative,” one person close to Windsurf told TechCrunch. Meaning it cost more to run the product than the startup could charge for it.
This is due to the high costs of using large language models (LLMs), the person explained. AI coding assistants are particularly pressured to always offer the most recent, most advanced, and most expensive LLMs because model makers are particularly fine-tuning their latest models for improvements in coding and related tasks like debugging.
This is a challenge compounded by fierce competition in the vibe-coding and code-assist market. Rivals include companies that already have huge customer bases like Anysphere’s Cursor and GitHub Copilot.
The most straightforward path to improving margins in this business involves the startups building their own models, thereby eliminating costs of paying suppliers like Anthropic and OpenAI.
“It’s a very expensive business to run if you’re not going to be in the model game,” said the person.
But that idea comes with its own risks. Windsurf’s co-founder and CEO, Varun Mohan, ultimately decided against the company building its own model — an expensive undertaking, the person said.
In addition, model makers are already competing directly. Anthropic offers Claude Code and OpenAI offers Codex, for instance.
Selling the business was a strategic move to lock in a high return before it could be undermined by the very companies that supplied its AI, including OpenAI and Anthropic, which were also entering the AI coding market.
Multiple people believe that the same pressure on margins Windsurf faced could be impacting Anysphere, the maker of Cursor, as well as vibe coders like Lovable, Replit, and others.
“Margins on all of the ‘code gen’ products are either neutral or negative. They’re absolutely abysmal,” said Nicholas Charriere, founder of Mocha, a vibe-coding startup and back-end hosting solution serving small and medium businesses (SMBs). He added that he believes the variable costs for all the startups in the sector are very close, likely within 10% to 15% of one another.
Unlike Windsurf, Anysphere has been growing so fast that it intends to remain an independent company, having already turned down acquisition offers, including, reports say, from OpenAI.
In addition to building a model, Anysphere could expect the cost of LLMs to decrease over time.
It’s not entirely clear how true that is. Rather than falling as expected, the cost of some of the latest AI models has risen, as they use more time and computational resources to handle complicated, multistep tasks.
When that will change remains to be seen. On Thursday, for instance, OpenAI introduced a new flagship model, GPT-5, with fees that are significantly less than its competitor, Anthropic’s Claude Opus 4.1. And Anysphere immediately offered this model as a choice for Cursor users.
Anysphere has also recently changed its pricing structure to pass along the increased costs of running Anthropic’s latest Claude model, particularly to its most active users. The move caught some of Cursor customers by surprise, since they didn’t expect additional charges on top of its $20-per-month Pro plan. Anysphere CEO Michael Truell later apologized for unclear communication about the pricing change in a blog post.
This is the rock and the hard place. Although Cursor is one of the most popular AI applications, having reached $500 million in ARR in June, the company’s user base may not be so loyal to the product if another company develops a tool that is superior to Cursor, investors say.
Beyond Cursor, other AI coding tools are also among the fastest growing startups of the LLM generation, like Replit, Lovable, and Bolt, and all of them rely on model makers as well.
Additionally, if this extremely popular business sector, already generating hundreds of millions in revenue or more a year, has difficulty building on top of model makers, what might it mean for other, more nascent industries?
Tomi Engdahl says:
Wall Street Journal:
An analysis of 96K public ChatGPT transcripts finds 100+ lengthy chats, of which dozens exhibited delusional traits where ChatGPT reinforced users’ fringe ideas — Online trove of archived conversations shows model repeatedly sending users down a rabbit hole of fringe theories about physics, aliens and the apocalypse
https://www.wsj.com/tech/ai/i-feel-like-im-going-crazy-chatgpt-fuels-delusional-spirals-ae5a51fc?st=Cqnx9E&reflink=desktopwebshare_permalink
Tomi Engdahl says:
Bloomberg:
Sources: German AI startup n8n is expected to raise hundreds of millions of euros led by Accel at a pre-money valuation of $2.3B, up from ~$350M four months ago — The venture capital firm Accel is leading a funding round for the German artificial intelligence startup n8n that would exponentially raise …
https://www.bloomberg.com/news/articles/2025-08-07/accel-leading-round-for-ai-startup-n8n-at-2-3-billion-valuation
Tomi Engdahl says:
Kalley Huang / The Information:
Source: Meta has acquired WaveForms AI, which is working on AI that understands and mimics emotion in audio and debuted in December with a $40M seed led by a16z — Meta Platforms has acquired WaveForms AI, a small startup working on artificial intelligence capable of understanding emotion …
https://www.theinformation.com/articles/meta-acquires-ai-audio-startup-waveforms
Tomi Engdahl says:
Ivan Mehta / TechCrunch:
The Browser Company launches a $20-per-month Dia Pro subscription plan, providing unlimited access to the web browser’s AI-powered chat and skills features
The Browser Company launches a $20 monthly subscription for its AI-powered browser
https://techcrunch.com/2025/08/06/the-browser-company-launches-a-20-monthly-subscription-for-its-ai-powered-browser/
The Browser Company has launched a Pro subscription plan for Dia, its new web browser that heavily integrates AI features. The plan costs $20 per month and provides unlimited access to Dia’s AI-powered chat and skills features.
The introduction of a paid tier means free users will now face usage limits on AI features. While The Browser Company hasn’t specified exact limits, CEO Josh Miller told The New York Times in July that the browser will remain free for those who use AI features “a few times a week.”
Miller also indicated that the startup plans to offer multiple subscription tiers, ranging from $5 per month to hundreds of dollars monthly. The current $20 plan appears to be one of several options based on different feature sets.
The Browser Company, which previously made the Arc browser, has raised $128 million from investors, including Pace Capital, Next Play Ventures, and notable tech execs like LinkedIn’s Jeff Weiner, Medium’s Ev Williams, Figma’s Dylan Field, Notion’s Akshay Kothari, and GitHub’s Jason Warner. This Pro plan represents the company’s first revenue-generating subscription service.
The startup faces growing competition when it comes to AI-enhanced browsers. Perplexity’s Comet browser is gaining steam, Opera is also prepping its Neon browser, and incumbents like Google and Microsoft are also integrating their AI assistants into their browsers.
Tomi Engdahl says:
Microsoft’s New Agentic Web Protocol Stumbles With Path Traversal Exploit
https://hackaday.com/2025/08/07/microsofts-new-agentic-web-protocol-stumbles-with-path-traversal-exploit/
If the term ‘NLWeb’ first brought to mind an image of a Dutch internet service provider, you’re probably not alone. What it actually is – or tries to become – is Microsoft’s vision of a parallel internet protocol using which website owners and application developers can integrate whatever LLM-based chatbot they desire. Unfortunately for Microsoft, the NLWeb protocol just suffered its first major security flaw.
The flaw is an absolute doozy, involving a basic path traversal vulnerability that allows an attacker to use appropriately formatted URLs to traverse the filesystem of the remote, LLM-hosting, system to extract keys and other sensitive information. Although Microsoft patched it already, no CVE was assigned, while raising the question of just how many more elementary bugs like this may be lurking in the protocol and associated software.
Microsoft’s plan to fix the web with AI has already hit an embarrassing security flaw
https://www.theverge.com/news/719617/microsoft-nlweb-security-flaw-agentic-web
This latest security issue highlights the challenges of security in an AI era.
Researchers have already found a critical vulnerability in the new NLWeb protocol Microsoft made a big deal about just a few months ago at Build. It’s a protocol that’s supposed to be “HTML for the Agentic Web,” offering ChatGPT-like search to any website or app. Discovery of the embarrassing security flaw comes in the early stages of Microsoft deploying NLWeb with customers like Shopify, Snowlake, and TripAdvisor.
The flaw allows any remote users to read sensitive files, including system configuration files and even OpenAI or Gemini API keys. What’s worse is that it’s a classic path traversal flaw, meaning it’s as easy to exploit as visiting a malformed URL. Microsoft has patched the flaw, but it raises questions about how something as basic as this wasn’t picked up in Microsoft’s big new focus on security.
“This case study serves as a critical reminder that as we build new AI-powered systems, we must re-evaluate the impact of classic vulnerabilities, which now have the potential to compromise not just servers, but the ‘brains’ of AI agents themselves,” says Aonan Guan, one of the security researchers (alongside Lei Wang) that reported the flaw to Microsoft. Guan is a senior cloud security engineer at Wyze (yes, that Wyze) but this research was conducted independently.
Microsoft’s plan to fix the web: letting every website run AI search for cheap
https://www.theverge.com/web/669437/nlweb-microsoft-ai-agents-open-web
NLWeb starts by offering ChatGPT-level search to any site or app, with just a few lines of code. It’s a new vision for the web.
Too much of that new communication, Guha thinks, is mediated by products like ChatGPT, Claude, and yes, even Bing. He doesn’t like the idea that the web will be utterly consumed by chatbots, which take all their knowledge and return no value. And he thinks he knows how to fix it.
Guha’s big idea is to make it easy for any website or app owner to add ChatGPT-style interaction features. With a few lines of NLWeb code, your choice of an AI model, and whatever data you supply to the model, you can have a custom chatbot up and running in just a few minutes. “It’s a protocol,” Guha says, “and the protocol is a way of asking a natural-language question, and the answer comes back in structured form.”
Tomi Engdahl says:
AI Price Gouging: Corporate Greed Is Out of Control
https://www.youtube.com/watch?v=cyMWig4AR4k
Corporations are “enhancing their pricing strategy” by combining AI with dynamic pricing. Delta, Walmart, Kroger, Wendy’s and other major corporations are using artificial intelligence to set prices based on data they’ve collected from you, effectively price gouging each of us on an individual basis. From Delta’s “full reengineering” of airline pricing to Kroger’s pilot program with facial recognition displays, the evidence is disturbing.
#Delta #Walmart
Chapters:
00:00 Intro
00:52 Uber Surge Pricing
01:58 Dynamic Pricing Isn’t New, But AI Is
02:46 Online Age Verification Gives More Data to Big Tech
03:25 AI Could Access Public and Private Data Records
04:11 Delta AI Dynamic Pricing
04:33 Delta Earnings Call
04:58 Fetcherr AI and Delta Receiving Backlash
06:30 Walmart Digital Displays
07:02 Walmart Dynamic Pricing
07:29 Inflated Grocery Prices
08:45 The Technology Could Be Used for Good
09:07 Lower Income Households Pay More for Internet
09:31 Digital Displays Hide Tariff Costs
10:27 Kroger Technology and Data
11:10 Kroger Digital Displays
12:07 Kroger Piloted Cameras and Facial Recognition
12:41 Kroger Surveillance Tech and Robotics
13:41 Wendy’s Dynamic Pricing
13:56 Wendy’s Earnings Call – AI
14:11 Wendy’s Digital Menu Boards
15:10 Wendy’s Earnings Call – Increase Revenue
16:16 Is This Legal?
Tomi Engdahl says:
Zac Hall / 9to5Mac:
Apple says Apple Intelligence will use OpenAI’s GPT-5 on iOS 26, iPadOS 26, and macOS Tahoe 26, with the system updates expected to arrive in September
Here’s when ChatGPT integration within Apple Intelligence will use GPT-5
https://9to5mac.com/2025/08/07/apple-intelligence-gpt-5-chatgpt-integration/
Tomi Engdahl says:
Zico Ghosh / The Hollywood Reporter:
A look at the impact of AI voice cloning on India’s dubbing and voiceover sector, as artists demand consent, credit, and fair pay amid a lack of laws for AI use
https://www.hollywoodreporter.com/movies/movie-news/ai-is-replacing-voice-artists-in-india-1236335714/
Tomi Engdahl says:
Jay Peters / The Verge:
During its GPT-5 livestream, OpenAI showed two charts that had scales all over the place, with Sam Altman later calling one “a mega chart screwup from us” — CEO Sam Altman called a strange graph in its GPT-5 presentation a ‘mega chart screwup.’
OpenAI gets caught vibe graphing
CEO Sam Altman called a strange graph in its GPT-5 presentation a ‘mega chart screwup.’
https://www.theverge.com/news/756444/openai-gpt-5-vibe-graphing-chart-crime
Tomi Engdahl says:
Edward Ludlow / Bloomberg:
Sources: Tesla is disbanding the Dojo supercomputer team, and its leader is leaving, after ~20 team members joined DensityAI, a new data center services startup — Tesla Inc. is disbanding its Dojo supercomputer team and its leader will depart the company, according to people familiar with the matter …
Tesla Disbands Dojo Supercomputer Team in Blow to AI Effort
https://www.bloomberg.com/news/articles/2025-08-07/tesla-disbands-dojo-supercomputer-team-in-blow-to-ai-effort
Tomi Engdahl says:
Meghan Bobrowsky / Wall Street Journal:
Sources: TBD Lab, a team under Meta Superintelligence Labs that houses many researchers poached from rival labs, is spearheading work on Llama’s newest version — Group is spearheading work on the latest version of Llama, the large language model that is Meta’s answer to ChatGPT
Meta’s Superintelligence AI SWAT Team Is Now Called TBD Lab
Group is spearheading work on the latest version of Llama, the large language model that is Meta’s answer to ChatGPT
https://www.wsj.com/tech/ai/meta-ai-superintelligence-team-6415a4f4?st=3Wxnq7&reflink=desktopwebshare_permalink
About one month in, Meta Platforms’ much-hyped new team devoted to building machine superintelligence has a name, an initial project—and a fast-growing list of employees poached from other artificial-intelligence labs.
Earlier this summer, the company placed all of its AI efforts under a new umbrella group, Meta META -1.32%decrease; red down pointing triangle
Superintelligence Labs. Charged with a mission to “bring personal superintelligence to everyone,” in the words of Chief Executive Mark Zuckerberg, MSL is overseen by Chief AI Officer Alexandr Wang, who joined the company as part of a $14 billion deal for a stake in his former startup, Scale AI.
At the forefront of the push to build a computer mind smarter than any human’s is a team dubbed TBD Lab, which houses many of the researchers the company has lured away from rival labs, in some cases with pay packages of tens or hundreds of millions of dollars.
TBD Lab, as in “to be determined,” is spearheading work on the newest version of Llama, the company’s large language model, according to people familiar with the matter.
Last week, Wang sent a memo to employees that was viewed by The Wall Street Journal. Wang wrote that TBD Lab would be working alongside Meta’s other AI teams on a variety of projects, including coming model releases, the extension of models’ reasoning capabilities and development of AI agents.
“Already in the past month, I’ve seen meaningful progress in each of these collaborations,” he wrote in the memo. “This enables us to be more technically ambitious, parallelize across several separate efforts and ultimately achieve frontier results more quickly.”
Tomi Engdahl says:
Stephanie Palazzolo / The Information:
Source: OpenAI awards bonuses to ~1,000 research and engineering employees, or ~30% of its staff, ranging from several hundred thousand to millions of dollars
OpenAI Pays Bonuses Ranging Up To Millions of Dollars to 1,000 Researchers, Engineers
https://www.theinformation.com/briefings/openai-pays-bonuses-ranging-millions-dollars-1-000-researchers-engineers
Tomi Engdahl says:
Duncan Riley / SiliconANGLE:
Decart, which offers real-time generative video and GPU optimization tech to cloud providers and AI companies, raised a $100M Series B at a $3.1B valuation
Decart raises $100M on $3.1B valuation to grow real-time AI video platform
https://siliconangle.com/2025/08/07/decart-raises-100m-3-1b-valuation-grow-real-time-ai-video-platform/
Artificial intelligence developer Decart.AI Inc. today announced that it has raised $100 million in new funding on a $3.1 billion valuation to expand research and development, scale up its infrastructure platform and accelerate go-to-market efforts for its real-time generative AI technologies.
Founded in 2023, Decart specializes in real-time generative video technology and graphics processing unit optimization. Its mission is to make AI fast, responsive and affordable enough to power dynamic, interactive experiences for millions of concurrent users.
Decart’s offering differs from traditional generative AI tools that produce static outputs by focusing on production-ready systems that can operate at low latency and high frame rates. The company says the output enables entirely new categories of applications in gaming, entertainment, robotics and beyond.
The company’s capabilities are focused on a GPU optimization stack that it says drastically reduces the cost of generating video content using diffusion models, from hundreds or even thousands of dollars per hour to under 25 cents. The cost breakthrough, it says, makes large-scale, real-time AI video generation not only technically feasible but also commercially viable.
The company offers two products, Oasis and MirageLSD. Oasis, launched in November 2024, is a real-time video model that reached more than a million users within three days of release. MirageLSD, which launched this week, builds on the momentum from Oasis with enhanced capabilities for high-quality, low-latency video transformation.
Tomi Engdahl says:
FUN
Toxic CEO: Replacing Everyone with AI
https://www.youtube.com/watch?v=THfBccihkVQ
Toxic CEO is back! This time, he’s got a genius plan to boost productivity: fire everyone and replace the office with AI.
Things go smoothly until the AI starts doing calculations and makes a cost-cutting suggestion no one saw coming (except maybe all of us).
Tomi Engdahl says:
The AI writing epidemic
https://youtu.be/y7pgdDBu8z4?si=cLKM-gjaxCUeuPAD
Tomi Engdahl says:
Luc Olinga / Gizmodo:
Sam Altman says OpenAI is restoring GPT-4o to ChatGPT and raising reasoning model limits for free and Plus users, as usage of reasoning models increases
OpenAI Brings Back Fan-Favorite GPT-4o After a Massive User Revolt
The move is a stunning reversal, proving that even the most powerful AI company can’t ignore a mutiny from its loyal user base.
https://gizmodo.com/openai-brings-back-fan-favorite-gpt-4o-after-a-massive-user-revolt-2000641214
After a disastrous 72 hours that saw its most loyal users in open revolt, OpenAI is making a major U-turn.
In a series of posts on X (formerly Twitter) Sunday, CEO Sam Altman announced that the company is bringing back its beloved older AI models, including GPT-4o, and dramatically increasing usage limits for paying subscribers, a clear peace offering to a furious customer base.
Tomi Engdahl says:
Gary Marcus / Marcus on AI:
GPT-5′s release was underwhelming, offering incremental improvements and failing to meet expectations, showing that pure scaling simply isn’t the path to AGI — A new release botched … and new research paper that spells trouble — GenerativeAI had a truly bad week.
GPT-5: Overdue, overhyped and underwhelming. And that’s not the worst of it.
A new release botched … and new research paper that spells trouble
https://garymarcus.substack.com/p/gpt-5-overdue-overhyped-and-underwhelming
Tomi Engdahl says:
Alex Kantrowitz / Big Technology:
Q&A with OpenAI COO Brad Lightcap on GPT-5, its dynamic reasoning, defining AGI, scaling vs. post-training, hallucinations, enterprise adoption, and more
OpenAI COO Brad Lightcap: GPT-5′s Capabilities, Why It Matters, and Where AI Goes Next
A conversation about what OpenAI’s long-anticipated flagship model tells us about different forms intelligence and AI’s trajectory from here.
https://www.bigtechnology.com/p/799049c8-5054-45c0-8ee7-9de1f2191759?postPreview=paid&updated=2025-08-08T15%3A09%3A29.180Z&audience=only_paid&free_preview=false&freemail=