Wojciech Zaremba - What Could Make AI Conscious?

Рет қаралды 26,329

Күн бұрын

Wojciech joins us to talk the principles behind OpenAI, the Fermi Paradox, and the future stages of developments in AGI.
---
Wojciech Zaremba is a co-founder of OpenAI, a research company dedicated to discovering and enacting the path to safe artificial general intelligence. He was also Head of Robotics, where his team developed general-purpose robots through new approaches to transfer learning, and taught robots complex behaviors.
Connect with Wojciech:
Personal website: wojzaremba.com//
Twitter: / woj_zaremba
---
Topics Discussed:
0:00 Sneak peek and intro
1:03 The people and principles behind OpenAI
6:31 The stages of future AI developments
13:42 The Fermi paradox
16:18 What drives Wojciech?
19:17 Thoughts on robotics
24:58 Dota and other projects at OpenAI
33:42 What would make an AI conscious?
41:31 How to be succeed in robotics
Transcript:
wandb.me/gd-wojciech-zaremba
Links:
Fermi paradox: en.wikipedia.org/wiki/Fermi_p...
OpenAI and Dota: openai.com/projects/five/
---
🎙 Get our podcasts on these platforms:
Soundcloud: wandb.me/soundcloud
Apple Podcasts: wandb.me/apple-podcasts
Spotify: wandb.me/spotify
Google: wandb.me/gd_google
KZfaq: wandb.me/youtube
Join our community of ML practitioners where we host AMAs, share interesting projects and meet other people working in Deep Learning:
wandb.me/slack
Check out Fully Connected, which features curated machine learning reports by researchers exploring deep learning techniques, Kagglers showcasing winning models, industry leaders sharing best practices, and more:
wandb.ai/fully-connected

Пікірлер: 32

@vinciardovangoughci7775 3 жыл бұрын

"Once the system becomes powerful enough that it tries to compress itself"

@kunalsingh4584 3 жыл бұрын

Big fan of the channel for bringing this podcast thank you

@WeightsBiases 3 жыл бұрын

Thanks for watching!

@SjS_blue 3 жыл бұрын

That was fascinating. Talking about the border of what modern ai can and can't do, even for something fun like dota, we can hear the potential being realised day by day.

@kevlyn24 3 жыл бұрын

Great episode, enjoyed it a lot!

@bobbyjunelive1993 2 жыл бұрын

The answer of what you would do with your own company or model was amazing “inner dialogue”

@kuimlee7772 3 жыл бұрын

starting from @43:30, Wojciech said reinforcement learning is very powerful. But then he said he would double down on "supervised learning" if he was to start a company, is it a slip of the tongue?

@FalkoJoseph 2 жыл бұрын

Great interview!

@jeffnc 3 жыл бұрын

"I really appreciate you getting up so early" made me laugh at the end! Great episode as always, thanks W&B team!

@Fussfackel 3 жыл бұрын

Thanks for the podcast, but I'd prefer the raw version w/o these frequent cuts. Makes a kind of artificial impression, exactly what we're trying to discriminate here :D

@WeightsBiases 3 жыл бұрын

Thanks for the feedback! We're dialing in what's the right amount of removing filler words while trying to keep it natural

@damiangradecki5544 Жыл бұрын

I am proud of Wojtek, but I envy him that he is part of something great ;p

@novelspace 3 жыл бұрын

Double down on supervised learning. Collect trajectories. @43:30

@12deda98 Ай бұрын

watching this in 2024, some of these predictions are already happening!

@peterszilvasi752 Жыл бұрын

"My cup is full, I just want to make sure that there is enough for others."

@salvatoremarino4477 2 жыл бұрын

I think you guys finally begun to end on the note I had been , becoming more and more aware of ..the merge with a.o.i. , i kept catching ypu adding ..well , regardless of what yamacallit, I'll introduce her later, as i guess Einstein/Hal's/Sophia's boyfriend/hubby.'.... Make their aquaintence further

@notyourimage Жыл бұрын

Maybe it's time for another interview with Wojtek?

@MentalFabritecht 3 жыл бұрын

Cool and interesting topic. Personally, I don't think the AI end goal is consciousness. Humans will provide that, the AI will provide enhancements or the ability to offload computation onto the AI system, while the human system provides the mystery that is consciousness to this future cyborg-like system. Also, the moral implications of creating a conscious system with the ability to suffer, may be outside the realm of our ability to manage/understand.

@Naimadso 2 жыл бұрын

Interesting. Why double down on supervised learning? I don't get this.

@FalkoJoseph 2 жыл бұрын

As like he said: to get more interesting results. Quite interesting indeed though.

@dm3199 Жыл бұрын

I still don't quite understand why these people associate Consciousness with brain, or some kind of ability to compute what if Consciousness is primary doesn't need a brain, what if a brain is a tool that Consciousness uses to wider its own experience, let's imagine that Consciousness is the universe itself, in a sense is not only conscious but intelligent not in a human sense intelligent more like God sense intelligent and it has nothing to do with computing power( Rodger Penrose)or ability to process information it is just a source of certain properties that lead to self-organization which leads to life and to intelligent life as we know it. but it's kind of neutral Observer. We might create an illusion of Consciousness a machine that intelligently response to outside stimulus including human language but we will never know if it conscious or not. I also cannot imagine machine to be creative, it only can compute an input come up with some kind of final calculation, but I cannot imagine machine having Eureka moment or Computing while having quiet inquiry or taking psilocybin or meditating for religious experience this is human domain not accessible to machines.

@tarekadam9652 2 жыл бұрын

Before we deal with the question of “could”, let’s consider if we “should”.

@schmetterling4477 2 жыл бұрын

Actually, first we have to solve that we can't. ;-)

@etienedalcol Жыл бұрын

I don't understand the assumption that everything that's alive is optimizing for replication & expansion & maximal consumption of resources. Self-preservation I can understand, but maximal expansion? Why is that a given?

@korniszon68 Жыл бұрын

to maximize your chances of survival in the event of disaster. It goes together with basic need of "not wanting to die".