Sam Altman

AI summary of what Sam Altman said

Sam Altman explains why he thinks AI can revolutionize Alzheimer's research, announcing his foundation's $100 million pledge to the cause. He argues that recent AI systems are smart enough to accelerate science and discover new knowledge on their own. He also addresses fears about AI, saying society will have to evolve and that such power should not rest in one person's hands.

• He says AI is good at synthesizing huge amounts of data across many fields, which could help crack the complexity of Alzheimer's disease.
• He says recent AI systems are discovering new knowledge on their own for the first time, which scientists say will make science move faster.
• He says AI should be put in people's hands early and often so society can build on it, acknowledging it will be messy and some jobs will be taken over entirely.

AI summary of what Sam Altman said

Sam Altman, co-founder of OpenAI, appears as a guest alongside Greg Brockman on the Core Memory podcast. He discusses OpenAI's mission of ensuring AGI benefits all of humanity, the importance of iterative deployment and a different framing of safety, and how shipping products like ChatGPT convinced the world AI was real. He also covers compute strategy, inequality scenarios, relations with the US government, and the upcoming Elon Musk lawsuit.

• He says OpenAI's launch of ChatGPT was the moment the world collectively accepted AI was real, and that shipping great, delightful products does more to convince people than proclamations about superintelligence.
• He says one of OpenAI's greatest contributions has been finding a different way to talk about safety through iterative deployment, with products getting safer as stakes rise, against pressure to do otherwise.
• He says the coming Musk trial is a real opportunity to tell OpenAI's story, and that the historical breaking point with Elon was his demand for absolute control of the company.

AI summary of what Sam Altman said

Sam Altman discusses the need for proof of human identity in an internet increasingly filled with AI-generated content and agents. He explains the origins and goals of World and World ID as a privacy-preserving way to verify that someone is a real person. He also says AI will lift everything up, including blockchain, and that agents will need ways to pay each other and move money around.

• He says AI has advanced faster than he expected, making it hard to tell whether online comments and interactions come from humans or AI.
• He explains that World ID is an effort to create a privacy-preserving proof of human for the internet, so the world can remain about people even as AI agents proliferate.
• He says the core ideas of World ID have stayed the same, though the specific ways it will matter, such as language models, were not obvious eight years ago.

AI summary of what Sam Altman said

Sam Altman says OpenAI released its blueprint on superintelligence now because the rate of progress is accelerating and capable models are coming soon, and the public needs time to debate ideas before decisions must be made. He argues resilience to AI risks like cyber and bio threats requires an emergent, society-wide response, not just one company's safety measures. He also says democratizing AI means both broad access to compute and giving people a voice in where it goes, and that making AI abundant and cheap, like electricity, is the best long-term strategy.

• He says OpenAI published the blueprint now because progress is accelerating and the public and leaders need time to debate ideas before major decisions arrive.
• He describes resilience as requiring society-wide defenses, using AI to defend against cyber and bio threats rather than relying on one company's safety measures alone.
• He says the best democratization strategy is building so much AI infrastructure that it becomes abundant and cheap, comparing it to how cheap energy lifted quality of life.

AI summary of what Sam Altman said

Sam Altman says he is optimistic about his kids growing up in a world where computers can do anything they ask, and believes they will have more agency than previous generations. He says OpenAI is concentrating its compute on accelerating science and the economy with its next set of models, and that AGI feels very close, with continuous learning or long-term memory as the main missing piece. He also welcomes the world finally taking seriously that the social contract will have to evolve.

• Sam Altman says it is awesome that kids will grow up in a 'Star Trek world' where they ask a computer for anything, and he trusts his kids to adapt and achieve amazing things with more agency than previous generations.
• Sam Altman says ARC-AGI 3 and other real, unsaturated benchmarks matter to OpenAI, but notes that AGI is hard to define and goalposts keep moving.
• Sam Altman says OpenAI wants to concentrate its compute on its most promising direction, comparing putting a million GPUs toward curing cancer versus making videos, and he would pick cancer research.

AI summary of what Sam Altman said

In this interview, Sam Altman says AI models have crossed into major economic utility, most visibly in coding and science, and that AI systems will soon handle multi-day and multi-week tasks and act like trusted senior employees. He discusses OpenAI's massive infrastructure spending, Stargate data centers, a new inference-only chip, and a partnership with building trades unions to train skilled construction workers.

• He says AI models have recently crossed a threshold into major economic utility, most noticeably in coding but also in science and knowledge work.
• He says OpenAI's business will look like selling tokens, with intelligence becoming a utility like electricity or water.
• He says OpenAI is building an inference-only chip focused on being the cheapest and most efficient per watt, with first chips deployed at scale by the end of the year.

AI summary of what Sam Altman said

Sam Altman, interviewed at TreeHacks on the Stanford campus, reminisces about his student days, dropping out to do a startup through Y Combinator, and the founding of OpenAI as an unpopular bet on scaling deep learning. He predicts current sophomores will graduate into a world with AGI, discusses coding agents as the next ChatGPT moment, and answers audience questions about OpenAI's spending, his happiness, and custom silicon.

• He recounts his Stanford years, including steam tunneling that led to a professor's car falling into an open manhole, and says building projects with friends mattered more than clubs.
• He describes OpenAI's founding as an unpopular bet on scaling deep learning, with the early team unsure it would ever be anything but a research lab.
• He says current students will graduate into a world with AGI, that coding agents were the next ChatGPT moment, and that OpenAI is building custom inference silicon while training on Nvidia.

AI summary of what Sam Altman said

Sam Altman, speaking to students at IIT Delhi, says no one should be purely excited or purely fearful about AI, and that its impact on jobs is uncertain even to OpenAI. He says some jobs will vanish, many will change less than expected, and new kinds of work will emerge, though the transition may be painful. He advises young people not to trust older generations' predictions and to develop their own intuitions about a rapidly changing world.

• He says people cannot be purely excited or purely fearful about AI, but must sit somewhere on a spectrum between the two.
• He says some jobs will disappear, many will change less than expected, and history shows new jobs emerge faster and stranger than predicted.
• He tells students not to trust him or their parents as predictors of the future, saying young people must quickly develop and trust their own intuitions.