Seonglae ChoAI progress is constrained by bottlenecks in essential economic sectors and steps of the innovation process, and that breakthroughs in manipulating the physical world are necessary
The comparison between AI and the dot com bubble is not entirely accurate due to distinct differences between the two.
Business standard
- Can it be better than the SOTA AI model (competitiveness)?
- Are customers willing to pay?
AI Companies
OpenAI
Deepmind
Anthropic AI
Meta FAIR
DeepSeek
Mistral AI
AllenAI
X.AI
Cohere
Together AI
Inflection AI
EleutherAI
MosaicML
Liquid AI
Nous Research
Sakana AI
Upstage AI
Neural Magic
Adept ai
Jina AI
Scatter Lab
CarperAI
inworld
Tiny corp
Scale.com
Kyutai AI
01.AI
Sarvam AI
Aleph Alpha
Abacus AI
Reka AI
Borealis AI
Rhymes AI
Prime Intellect
World Lab
GoodFire AI
Isomorphic Labs
Amazon AGI Lab
AI2
Enigma AI
Scale AI
Thinking Machines
Menlo
Moonshot AI
AI Vision Companies
AI Industry Notion
Automation + Augmentation
In practice, AI is an overlay to technology. It abstracts technology and makes it easy to use.
Top Companies
The One AI to Rule Them All
Trying to grok the outlook for general-purpose consumer AI...
https://spyglass.org/consumer-ai-services-apps/
Serious problem: A Few Influencers Are Shaping AI
A Few Social Media Influencers Are Shaping AI
Tweets lead to attention, which leads to higher citation numbers
https://spectrum.ieee.org/social-media-ai
Industry Landscape
The Anthropic Economic Index
Announcing our new Economic Index, and the first results on AI use in the economy
https://www.anthropic.com/news/the-anthropic-economic-index
The One-person Billion-dollar Company
Can AI agents make you a billionaire?
https://every.to/napkin-math/the-one-person-billion-dollar-company
Why transformative AI is really, really hard to achieve
A collection of the best technical, social, and economic arguments
https://zhengdongwang.com/2023/06/27/why-transformative-ai-is-really-really-hard-to-achieve.html
AI Will Create More Developers, Not Less
There are way more technologies that ought to be built that aren’t.
https://interconnect.substack.com/p/ai-will-create-more-developers-not
More complex tasks show greater speed improvements
- High school level tasks: approximately 9x faster, college level tasks: approximately 12x faster.
- Success rates slightly decrease with complexity (college-level task success rate: 66%).
Differences in usage patterns by country
- Low-income countries: higher proportion of educational use
- High-income countries: higher proportion of work and personal use
Impact varies significantly by occupation
- Simple 'frequency of use' is less important than actual impact considering success rate and time required.
- Data entry and radiology show overestimated impact, while teachers and software developers show relatively underestimated impact.
Augmentation still exceeds automation
- Currently augmentation 52% / automation 45%
- Automation share is gradually increasing over time.
Anthropic Economic Index: new building blocks for understanding AI use
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
https://www.anthropic.com/research/economic-index-primitives
Model Market Fit
Model-Market Fit
Product-market fit has a prerequisite that most AI founders ignore. Before the market can pull your product, the model must be capable of doing the job. That's Model-Market Fit.
https://www.nicolasbustamante.com/p/model-market-fit
