Latest news with #GregBrockman
Mint
24-04-2025
- Business
- Mint
The agentic AI revolution isn't the future, it's already here
Jensen Huang unveiled a simple exponential curve at the Consumer Electronics Show in January, where he talked about how AI has moved from Perception AI (machine learning, speech recognition, etc) to Generative AI (with ChatGPT and other models), and how AI will then become Agentic (a coding and personal assistant), before moving to Physical AI with robots and self-driving cars . The last two years have seen startling advances in GenAI, but the phrase 'AI Agent' has stayed mostly on whiteboards for years, invoking a future in which AI does not merely tell you what to do, but goes out into the wide world to do it. In technology, as in life, things happen gradually and then suddenly. The 'sudden' moment for AI agents seems to have come this week, as the future tipped into the present, headlined by the release of the confusingly named o3 and o4-mini by ChatGPT-maker OpenAI. Many of us, however, are yet to notice this tipping point. But first, what exactly is an AI agent? In classic computer-science terms, an agent is a program that perceives its environment, decides what to do next and then acts towards a goal—iteratively, autonomously and with minimal human micro-management. Chatbots, by contrast, are largely reactive: they return text when prompted, but rarely take the next step. Agents blur that boundary, wielding tools such as browsers, code, CRM dashboards or payment interfaces to complete multistep tasks that once required a person's cursor and keyboard. So, a chatbot will help plan your perfect holiday itinerary; an AI Agent will also do the bookings for you. Even before this month, there were proto agents out there. Deep research by almost every large player from OpenAI to Google could compile sources or summarize regulations to produce a research report on almost anything. OpenAI's browser agent Operator took this further, piloting its own mini-browser to read, scroll and fill forms online. Adept's ACT1 learnt to operate enterprise SaaS apps; and so on. But, arguably, the biggest leap came when OpenAI unveiled o3 and its big brother o4. These were not mere models, but as OpenAI President Greg Brockman said, '…they're really AI systems." That distinction matters. A model is the neural network brain that predicts the next token; a system wraps that brain in orchestration layers—tool routers, memory stores, validators, even secondary models—that monitor progress and decide which tool (or subskill) to invoke next. In other words, o3/o4 are agentic by design: they can write code, run it, open a browser to verify an answer and loop until a goal is satisfied, all behind a deceptively simple chat interface. For end users, it still feels like chatting; under the hood, the system runs through a checklist the way an assistant or intern might. OpenAI is not alone. Google's Gemini is folding agent capabilities into Workspace; Microsoft's Copilot is adding 'autocomplete tasks' for Excel and Teams; and Salesforce is training AI representatives that log calls automatically. It is agents that will finally bring GenAI to enterprises. Finance teams are trying out agents that monitor cash flow anomalies at 3am and file ledger fixes before humans log in. E-commerce firms are letting agents test storefront copy and take the better one live. In manufacturing, prototype maintenance agents read sensor data, cross-check manuals and schedule down time. Professional services giants are piloting 'first draft' research partners that read contracts, assemble evidence and even initiate further information requests. Agents will transform service-as-a-software, with users paying for services delivered by orchestrated agents rather than for access to software. Agents raise ethical questions. There is something deeply uncomfortable about giving human agency to a machine. If models hallucinate, multi-agent frameworks will hallucinate in a compounded way, resulting in a high probability of task failure. Above all, agents are a direct threat to human jobs. Sam Altman predicts agents 'will materially change the output of companies," while Bill Gates calls them a 'shock wave" that turns apps into proactive assistants. Huang predicts that 'IT will become the HR of AI agents" as the CIO decides which agents to bring into the workplace to augment or replace human resources. As I conclude the last of my Tech Whisperer columns for Mint , here's a thought: The companies that figure out how—and how far—to delegate work to agent colleagues will set the pace for the rest of the decade. Everyone else will be playing catch-up. The author is a founder of AI&Beyond and the author of 'The Tech Whisperer'.
Time of India
21-04-2025
- Health
- Time of India
AI's medical hits and misses: Some patients get relief from years of suffering; others are misdiagnosed
OpenAI president and cofounder Greg Brockman has claimed that artificial intelligence is beginning to make a meaningful difference in people's lives, particularly in areas such as medical diagnoses. In a recent post on X, Brockman said, 'I'm hearing more and more stories of ChatGPT helping people fix longstanding health issues.' He followed up with an anecdote involving a ChatGPT user, who had suffered from chronic back pain for over a decade. Despite trying physiotherapy and other treatments, relief had remained elusive. — gdb (@gdb) With all else failing, the individual fed detailed information into ChatGPT — including history, pain triggers, and exercises tried. The user claimed this led to a breakthrough, with pain levels decreasing by 60–70%. Not just ChatGPT Back in November, Elon Musk 's AI chatbot Grok made headlines for similar reasons. Users had begun uploading medical scans, including MRIs and X-rays, seeking diagnostic insights. Discover the stories of your interest Blockchain 5 Stories Cyber-safety 7 Stories Fintech 9 Stories E-comm 9 Stories ML 8 Stories Edtech 6 Stories Musk had encouraged this and urged users to 'try submitting x-ray, PET, MRI, or other medical images to Grok for analysis', adding that the tool 'is already quite accurate and will become extremely good'. Some reported helpful feedback. However, others were misdiagnosed, highlighting the risks of relying solely on AI for medical interpretation. Promise vs precision The role of AI in healthcare remains a widely debated topic, raising questions about its potential and its pitfalls. A study led by Dr Hirotaka Takita and Associate Professor Daiju Ueda at Osaka Metropolitan University's Graduate School of Medicine explored the diagnostic performance of generative AI. Their research, reported by IANS, found that the average diagnostic accuracy was 52.1%, with newer models performing on par with non-specialist doctors. However, specialists still outperformed AI significantly, maintaining a 15.8% higher diagnostic accuracy. Meanwhile, a Reuters-reported study revealed troubling disparities. AI systems were shown to recommend different treatment paths for the same condition based purely on a patient's socioeconomic and demographic profile. For example, advanced tests such as CT scans or MRIs were more often suggested for high-income patients, while low-income patients were more frequently advised against further testing — a mirror to the current inequalities in healthcare. On the other hand, in October last year, ET reported that Mumbai-based AI startup successfully assisted in diagnosing tuberculosis in a patient, whose symptoms had confused several doctors. Experts agree that AI has a role to play in assisting medical professionals, but its impact hinges on the quality and diversity of the data it is trained on. Caution is advised when using AI for self-diagnosis. 'Imperfect answers might be okay for people purely experimenting with the tool, but getting faulty health information could lead to tests or other costly care you don't actually need,' said Suchi Saria, director of the machine learning and healthcare lab at Johns Hopkins University.
Yahoo
28-02-2025
- Business
- Yahoo
Energy Department scientists participate in ‘AI Jam' with OpenAI, Anthropic
Nearly 1,000 scientists from multiple U.S. National Laboratories gathered Friday to test artificial intelligence (AI) models from leading firms like OpenAI and Anthropic in an effort to harness the advancing technology for science and national security purposes. The event, dubbed '1,000 Scientist AI Jam Session,' involved nine national labs from across the country in what OpenAI, the parent company of ChatGPT, called 'a first-of-its-kind' event. OpenAI organized the event as part of its ongoing collaboration with the Department of Energy, which was announced nearly a month ago by CEO and co-founder Sam Altman at an event in Washington. Part of the testing included on OpenAI's o3-mini, launched late last month. OpenAI President and co-founder Greg Brockman said the advancing of scientific research is 'one of the most promising applications of AI.' Multiple AI firms participated in the event Friday, including Anthropic, which launched its Claude 3.7 Sonnet model this week. Anthropic said the model, which is being tested by scientists at the AI Jam, is the industry's first 'hybrid AI reasoning model,' meaning it can provide quick responses or extended, more detailed answers with more 'thinking' time. Scientists were expected to test a variety of Claude's capabilities, including experiment planning, problem understanding, literature search, code generation and result analysis, the AI company said in a blog post Friday. The researchers will use real-world issue areas related to their respective fields. 'AI has the potential to dramatically accelerate scientific discovery and technological development, compressing decades of scientific progress into just a few years,' Anthropic wrote in the post. 'Through initiatives like the AI Jam session, we are demonstrating how industry and government can work together to harness AI's transformative potential while addressing potential risks through rigorous testing,' the post added. Anthropic also has an existing partnership with the DOE including the National Nuclear Security Administration. The firm collaborated with the agencies last April to evaluate a model in a classified environment to evaluate how large language models can contribute or help national security risks in the nuclear space. Energy Secretary Chris Wright, who joined Brockman at Oak Ridge National Laboratory on Friday, compared the collaboration to the Manhattan Project, writing, 'AI development is a race that the United States must win.' The event comes just weeks after OpenAI launched a new version of its popular ChatGPT model specifically tailored to government agencies and workers. Under ChatGPT Gov, federal agencies will have access to OpenAI's top models even when dealing with sensitive information. Copyright 2025 Nexstar Media, Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.
The Hill
28-02-2025
- Business
- The Hill
Energy Department scientists participate in ‘AI Jam' with OpenAI, Anthropic
Nearly 1,000 scientists from multiple U.S. National Laboratories gathered Friday to test artificial intelligence (AI) models from leading firms like OpenAI and Anthropic in an effort to harness the advancing technology for science and national security purposes. The event, dubbed ' 1,000 Scientist AI Jam Session,' involved nine national labs from across the country in what OpenAI, the parent company of ChatGPT, called 'a first-of-its-kind' event. OpenAI organized the event as part of its ongoing collaboration with the Department of Energy (DOE), which was announced nearly a month ago by CEO and co-founder Sam Altman at an event in Washington. Part of the testing included on OpenAI's o3-mini, launched late last month. OpenAI President and co-founder Greg Brockman said the advancing of scientific research is 'one of the most promising applications of AI.' Multiple AI firms participated in the event Friday, including Anthropic, which launched its Claude 3.7 Sonnet model this week. Anthropic said the model, which is being tested by scientists at the AI Jam, is the industry's first 'hybrid AI reasoning model,' meaning it can provide quick responses or extended, more detailed answers with more 'thinking' time. Scientists were expected to test a variety of Claude's capabilities, including experiment planning, problem understanding, literature search, code generation and result analysis, the AI company said in a blog post Friday. The researchers will use real-world issue areas related to their respective fields. 'AI has the potential to dramatically accelerate scientific discovery and technological development, compressing decades of scientific progress into just a few years,' Anthropic wrote in the post. 'Through initiatives like the AI Jam session, we are demonstrating how industry and government can work together to harness AI's transformative potential while addressing potential risks through rigorous testing,' the post added. Anthropic also has an existing partnership with the DOE including the National Nuclear Security Administration (NNSA). The firm collaborated with the agencies last April to evaluate a model in a classified environment to evaluate how large language models can contribute or help national security risks in the nuclear space. Energy Secretary Chris Wright, who joined Brockman at Oak Ridge National Laboratory on Friday, compared the collaboration to the Manhattan Project, writing, 'AI development is a race that the United States must win.' The event comes just weeks after OpenAI launched a new version of its popular ChatGPT model specifically tailored to government agencies and workers. Under ChatGPT Gov, federal agencies will have access to OpenAI's top models even when dealing with sensitive information.
