David Luan: Why Nvidia Will Enter the Model Space & Models Will Enter the Chip Space | E1169
ฝัง
- เผยแพร่เมื่อ 28 มิ.ย. 2024
- David Luan is the CEO and Co-Founder at Adept, a company building AI agents for knowledge workers. To date, David has raised over $400M for the company from Greylock, Andrej Karpathy, Scott Belsky, Nvidia, ServiceNow and WorkDay. Previously, he was VP of Engineering at OpenAI, overseeing research on language, supercomputing, RL, safety, and policy and where his teams shipped GPT, CLIP, and DALL-E. He led Google’s giant model efforts as a co-lead of Google Brain.
-----------------------------------------------
Timestamps:
(00:00) Intro
(01:03) Lessons from Google Brain & Their Influence on Building Adept
(05:06) Why It Took 6 Years for ChatGPT to Emerge After Transformers
(06:49) Takeaways from OpenAI
(09:57) The Key Bottleneck in AI Model Performance
(16:06) Understanding Minimum Viable Capability Levels & Model Scale
(20:17) The Future of the Foundational Model Layer
(33:26) Adept’s Focus for Vertical Integration for AI Agents
(35:53) The Distinction Between RPA & Agents
(40:24) The Co-pilot Approach: Incumbent Strategy or Innovation Catalyst
(42:46) Enterprise AI Adoption Budgets: Experimental vs. Core
(46:53) AI Services Providers vs. Actual Providers
(49:32) Open vs. Closed AI Systems for Crucial Decision Making
(54:18) Quick-Fire Round
-----------------------------------------------
In Today’s Episode with David Luan We Discuss:
1. The Biggest Lessons from OpenAI and Google Brain:
What did OpenAI realise that no one else did that allowed them to steal the show with ChatGPT?
Why did it take 6 years post the introduction of transformers for ChatGPT to be released?
What are 1-2 of David’s biggest lessons from his time leading teams at OpenAI and Google Brain?
2. Foundation Models: The Hard Truths:
Why does David strongly disagree that the performance of foundation models is at a stage of diminishing returns?
Why does David believe there will only be 5-7 foundation model providers? What will separate those who win vs those who do not?
Does David believe we are seeing the commoditization of foundation models?
How and when will we solve core problems of both reasoning and memory for foundation models?
3. Bunding vs Unbundling: Why Chips Are Coming for Models:
Why does David believe that Jensen and Nvidia have to move into the model layer to sustain their competitive advantage?
Why does David believe that the largest model providers have to make their own chips to make their business model sustainable?
What does David believe is the future of the chip and infrastructure layer?
4. The Application Layer: Why Everyone Will Have an Agent:
What is the difference between traditional RPA vs agents?
Why is agents a 1,000x larger business than RPA?
In a world where everyone has an agent, what does the future of work look like?
Why does David disagree with the notion of “selling the work” and not the tool?
What is the business model for the next generation of application layer AI companies?
-----------------------------------------------
Subscribe on Spotify:
open.spotify.com/show/3j2KMcZ...
Subscribe on Apple Podcasts:
podcasts.apple.com/us/podcast...
Follow Harry Stebbings on Twitter:
/ harrystebbings
Follow David Luan on Twitter:
/ jluan
Follow 20VC on Instagram:
/ 20vchq
Follow 20VC on TikTok:
/ 20vc_tok
Visit our Website:
www.20vc.com
Subscribe to our Newsletter:
www.thetwentyminutevc.com/con...
-----------------------------------------------
#20vc #harrystebbings #davidluan #adeptai #venturecapital #ai #openai #nvidia #deepmind #chatgpt #apple - วิทยาศาสตร์และเทคโนโลยี
Subscribe to the 20VC TH-cam channel for more great interviews: www.youtube.com/@20VC
David’s honest, well informed input is much appreciated.
Agreed. Well spoken too this was a great insight for someone who is outside looking in
This has been the most insightful look into the future I've seen in recent times since Carl Shulman's interview by Dwarkesh Patel
This was a video that I needed to see. The content is getting better and better, Harry!
This guy knows so much about so many things! Brilliant.
One of the best interviews I’ve seen in a long time. Both of you covered such broad and novel questions. David had such lucid, well-explained answers to Harry’s questions. Harry I love how good your questions were. All questions the audience would want to know the answers to. And your ability to peel the onion and keep diving down before changing the topic is phenomenal. You have a new subscriber!
Another gem from 20VC!
Very clear thinker! Amazing episode!
Apple will get their AI to learn live as users try to use it. It will fail but learn.
Harry you are a brilliant interviewer. Also David was a great guest. Dual effort. Thank you.
Ur a wizard harry !
Brilliant. If there was a button for "TH-cam - show me this first, always", Id tap it.
Excelent interview!!
The fast way to implement 99.999(1) % safety self-driving vehicles in a city begins:
1) setting a sector where Human drivers are not allowed to drive in
2) Set that sector with cameras, move detectors and space scanners, all of that for modeling in real time the sector, let's call that 'A' dynamical set
3) Make the self driving vehicles to receive data from A.
4) by the exchange of data between A and each vehicle, the system will deploy patterns of predictability of vehicles, traffic, and anticipation ... plus, each vehicle and the system 'knowing' where/when one of each one is going on and each destiny. (for each vehicle) just by a subset of A, its own data, and a subset of date of the vehicles that surrounds it is enough ... like birds in a flock ...
OK, pretty complicated (lot of subtasks don't exposed in the resume) and controversial(1) for westerners by denying human drivers in the sector ... but after observing the outcomes , see how traffic diminish , how people move faster and easier in the sector , how the vehicles move faster without crashing , and how the sector overall economy increase ... People would support to build new 'self-driving vehicles sectors'
(1) the risks would come from animals or humans throwing things or themselves to the streets with bad intentions and/or suicidal attempts (but that can be anticipated by extending A sector to spaces from A where the vehicles doesn't circulate. During Earthquakes, it seems recommended to bypass activity, open the vehicle's doors and allow passengers to escape.
(2) for preventing lags in the processes of development and implementation ... don't allow self driving vehicles out of the sector ...
(**) China! Are you ready ??
Not new tech deployments required , just implementation of what already you have.
This video has a very high ratio of insights per min!
He spent entire minutes just repeating obvious stuff.
An evolutionary learning algorithm that rewards efficient solutions, as it learns, is needed to save electricity and compute.
Amazing interview, great questions
Thank you!
This is so good!
It’s reasoning compute above base intelligence!
Just like what Alex said!❤
But we need the research phase to get transformers! The large problem solving teams only worked because of the research.
damn.. this guy is smart
With regards to T1 cloud providers needing to make models for the life of them to survive in the future... what is up with AWS... bedrock...? I guess they invested in Anthropic but... is that their business saving move? I would think they'd want to be more like Elon Musk and have their own brand of model like with xAI that's competing with the best.
I wrote a whole post about how deepmind solves unsolved problems only to go back and realize i misheard what you said in the intro.
What did you get wrong?
He got things wrong>>> Model performance in agents improves regardless of the Base level. You can see this in Agent Hospital. GPT4o can already be used to make 3 d printed objects just by using your voice. @David Luan He is a smart guy, but we can do better.
He says "like" way too much