T
OpenAI’s Sam Altman Talks ChatGPT, AI Agents and Superintelligence — Live at TED2025
@ TED
04/12/25
Takeaways
SA
Sam Altman
04/12/25
@ TED
It's an honor to be here at TED.
SA
Sam Altman
04/12/25
@ TED
The new image generation model is part of GPT-4o, integrating all the intelligence from the model, which allows it to perform impressively.
SA
Sam Altman
04/12/25
@ TED
You can either feel threatened by AI doing your job or see it as a new tool that enhances your capabilities, similar to past technological revolutions.
SA
Sam Altman
04/12/25
@ TED
The expectation for job performance will increase, but the capabilities provided by AI will allow people to meet those expectations easily.
SA
Sam Altman
04/12/25
@ TED
The creative spirit of humanity is crucial, and we aim to build tools that enhance creativity, enabling people to create better art and content.
SA
Sam Altman
04/12/25
@ TED
We need to establish a new model for the economics of creative output, especially as AI builds on the creativity of others.
SA
Sam Altman
04/12/25
@ TED
As access to creativity becomes democratized, new business models will emerge that we are excited to explore, although the specifics are still uncertain.
SA
Sam Altman
04/12/25
@ TED
If you want to generate art in the style of a living artist, our current model won't allow that, but we are exploring a new model where artists can opt in and receive revenue.
SA
Sam Altman
04/12/25
@ TED
Open source has an important role, and we are developing a powerful open-source model that will be better than existing ones.
SA
Sam Altman
04/12/25
@ TED
The growth of ChatGPT has been unprecedented, with 500 million weekly active users, and it's growing rapidly.
SA
Sam Altman
04/12/25
@ TED
AI is becoming an extension of ourselves, learning from our interactions over time to help us achieve our best.
SA
Sam Altman
04/12/25
@ TED
AI for science is a major focus, as it can significantly enhance productivity and lead to new scientific discoveries.
SA
Sam Altman
04/12/25
@ TED
We expect to see meaningful progress against diseases with AI-assisted tools in the near future.
SA
Sam Altman
04/12/25
@ TED
The process of creating software has already transformed significantly, and we anticipate another major shift in the coming months with agentic software engineering.
SA
Sam Altman
04/12/25
@ TED
There have been moments of awe regarding AI's potential, and I often wonder how far this will go and what it will become.
SA
Sam Altman
04/12/25
@ TED
I believe there will be powerful AI models that can be misused in significant ways, including new forms of bioterrorism and cybersecurity threats.
SA
Sam Altman
04/12/25
@ TED
We have a preparedness framework that outlines how we check for safety before releasing AI models.
SA
Sam Altman
04/12/25
@ TED
Our safety track record is strong, with about 10% of the world using our systems regularly, despite differing views on AI safety.
SA
Sam Altman
04/12/25
@ TED
As AI systems become more capable, we need to ensure that we have safety processes in place throughout their development, not just when they become powerful.
SA
Sam Altman
04/12/25
@ TED
The iterative process of deploying AI systems, gathering feedback, and learning is crucial for building safe systems as we advance into more complex AI.
SA
Sam Altman
04/12/25
@ TED
ChatGPT is not AGI because it doesn't continuously learn or improve, nor can it perform any knowledge work autonomously.
SA
Sam Altman
04/12/25
@ TED
Defining AGI is complex; if you ask ten OpenAI researchers, you'll get fourteen different definitions.
SA
Sam Altman
04/12/25
@ TED
The focus should shift from pinpointing an AGI moment to recognizing that AI will continue to evolve beyond our current understanding.
SA
Sam Altman
04/12/25
@ TED
We need to build a society that can harness the benefits of advanced AI while ensuring its safety as it surpasses human capabilities.
SA
Sam Altman
04/12/25
@ TED
The challenge with agentic AI is balancing its capabilities with necessary guardrails to prevent misuse.
SA
Sam Altman
04/12/25
@ TED
As with any new technology, it takes time for people to become comfortable with agentic AI, similar to how we adapted to online transactions.
SA
Sam Altman
04/12/25
@ TED
The stakes are higher when AI has access to sensitive information and systems, making safety a critical concern.
SA
Sam Altman
04/12/25
@ TED
We need a framework for external safety testing of advanced AI models to understand and mitigate potential dangers before they are released.
SA
Sam Altman
04/12/25
@ TED
It's essential to define rigorous testing for AI models to ensure society understands the risks associated with their deployment.
SA
Sam Altman
04/12/25
@ TED
I think like anyone else, I'm a nuanced character that doesn't reduce well to one dimension.
SA
Sam Altman
04/12/25
@ TED
Our goal at OpenAI is to make artificial general intelligence (AGI) and distribute it safely for the broad benefit of humanity.
SA
Sam Altman
04/12/25
@ TED
We've put incredibly capable AI with a high degree of safety in the hands of many people, enabling them to do amazing things.
SA
Sam Altman
04/12/25
@ TED
I think it's fair that we should be open sourcing more, as we now have a better understanding of the impact these systems can have.
SA
Sam Altman
04/12/25
@ TED
We are one player in this AI revolution, trying to steward this technology into the world responsibly, despite having made mistakes.
SA
Sam Altman
04/12/25
@ TED
Having a child has changed a lot for me; the cost of not being with my kid is incredibly high, but I always cared about not destroying the world.
SA
Sam Altman
04/12/25
@ TED
I think people slow things down all the time because the technology is not ready or safe enough, and there are efforts to pause or delay releases.
SA
Sam Altman
04/12/25
@ TED
I think the caricature of this as just a crazy race misses the nuance that people are trying to put out models quickly while caring about safety.
SA
Sam Altman
04/12/25
@ TED
I think it's very cool that AI can learn the collective value preferences of what everybody wants, rather than just a small elite making decisions.
SA
Sam Altman
04/12/25
@ TED
I believe AI can help us make better collective governance decisions than we could before, guiding us to consider different perspectives.
SA
Sam Altman
04/12/25
@ TED
I hope my kids will grow up in a world of incredible material abundance, where individual ability and impact far exceed what a person can do today.
SA
Sam Altman
04/12/25
@ TED
I think my kids will look back at us with pity and nostalgia, thinking we lived such limited lives in a world that sucked.