OpenAI's Greg Brockman: The Future of LLMs, Foundation & Generative Models (DALL·E 2 & GPT-3)

Greg Brockman, President and Co-Founder of @OpenAI, joins Alexandr Wang, CEO and Founder of Scale, to discuss the role of foundation models like GPT-3 and DALL·E 2 in research and in the enterprise. Foundation models make it possible to replace task-specific models with those that are generalized in nature and can be used for different tasks with minimal fine-tuning.
In January 2021, OpenAI introduced DALL·E, a text-to-image generation program. One year later, it introduced DALL·E 2, which generates more realistic, accurate, lower-latency images with four times greater resolution than its predecessor. At the same time, it released InstructGPT, a large language model (LLM) explicitly designed to follow instructions. InstructGPT makes it practical to leverage the OpenAI API to revise existing content, such as rewriting a paragraph of text or refactoring code.
Before creating OpenAI, Brockman was the CTO of Stripe, which he helped build from four to 250 employees. Watch this talk to learn how foundation models can help businesses benefit from applications that they can create more quickly than with past generations of AI tools.

[@mrmonkeboy]

its crazy that after two weeks this video has under 300 likes and 22 comments. Future humans will wonder why no one was listening to this new science. Future AIs will look back at the birth of their species, at their creators and become amazed that no one was listening.

[@TBOBrightonandHove]

Unprecedented times. This is just the beginning. The application I am looking forward to most (and which I believe has the potential to benefit every individual on this planet) is a personal Assistant/Tutor/Therapist/Friend. Her we come...

[@johnholdsworth1878]

I could listen to this guy all day

[@alphacore4332]

well done to scaleAI for generating such a high quality video - very realistic

[@vincentcummings1]

Singularity is Near, 2005, Greg mentions it at

[@simonliddell1897]

I love that comment that when scaling benefits start to decline that will be the great moment of opportunity. Just a passionate curiosity to what to dig into solving newer harder problems. Go Greg I'm sure the time will come but at the moment we're all enjoying the ride as the curve keeps on going !

[@alexdubois6585]

It seems that tooling and help for data analysis is what is keep to continuously develop and nurture in these teams. Time to train retrospection models.

[@MicahBratt]

Talk about efficiency. Being able to train a model which is then able to scale to everyone on the planet is pretty incredible.

[@Nattyog]

Great interview! Exciting times

[@jon7707]

Greg brockman the legend himself, cant believe it wow

[@jmarkinman]

Cool, can't wait for one shots on this.

[@LinfordMellony]

Great insights from both Greg and Alexander being in the AI tech industry. AIs will definitely have a boost given that chatbots are now able to solve or able to answer mathematical and programming language. There's still improvements to be made though on all models even image generators like Bluewillow. Can't wait to see different models scale up to cater big studies/fields.

[@sophontec2822]

Great job! Everything starts from the sentiment neuron paper.

[@vidbina]

kind of humbling to hear how even the best are challenged in "understanding" what these magic machines really do. Stoked to see what we'll learn in the area of "explainability" but even if that doesn't advance much, I'm stoked to see which assist functions will enter everyone's toolbox because of this tech. And yes, pls hijack the term web3 for this stuff. 🤣

[@jzimmer11]

... about Moore's Law. He forgot to mention that you can also build more specialised hardware even without increasing, e.g., the density of transistors.

[@web3global]

Amazing Conversation! 💯

[@deeplearningpartnership]

Great interview.

[@dAvrilthebear]

The first internet connection happened in 1969. By 1989 this technology was already being used by the largest businesses and banks, by 1999 the Internet had reached most businesses and started exploding to home users, and by 2009 we had most of the modern infrustructure for personal use (including the smart phone).

[@jamespercy8506]

not reinventing the wheel but not being beholden to past practices, whatever is relevant. Sounds like elegant common sense to me!

[@CB-fz3li]

'Our charter definition is outperform humans at most economically valuable work' - I am not entirely convinced that their stated desire for AI to have a positive impact on people really sits well with their charter.