I spent this week at the Jefferies Private Internet Conference in Santa Monica hearing some incredible talks on the future of #AI from people like Marc Andreessen of Andreessen Horowitz, Sarah Tavel of Benchmark , Adam D'Angelo of Quora and OpenAI, Neerav Kingsland of Anthropic and many others. #JefferiesPrivateInternet
Key Takeaways:
🌐In AI the biggest concern right now is concern itself - if fear prevails technological progress will grind to a halt. If that same fear was applied to electricity, the automobile, the microchip or the internet we would never have had those innovations.
📈It's important to remember we're on an exponential growth curve - GPT2 was basically useless and that was a couple year ago. Today's models have reduced 1 day of work to 30 seconds of work. The next models will reduce 1 year of work to 30s. Then 10 years of work to 30s. Then 100 years of work to 30s. In the next decade we will reduce 1,000 years of work down to 30s.
🏰We are witnessing an escalation in barriers to entry with training costs of todays best models sitting around $100 million, this will rise to $1 billion and up to $10 billion, ultimately paving the way for an oligopoly dominated by major players like OpenAI, Anthropic, and Google.
🛠️A robust second tier of models is likely to emerge, that although slightly behind the frontier models, maybe by 1 year or so, will play a vital role in a diversifying ecosystem. While the tier 1 models are likely to support incredible applications like replacing software engineers, the tier 2 models are likely to commoditize and will support software application layer business apps. Both can be profitable.
⚔️The current political landscape reflects a dichotomy in government attitudes towards innovation: while showing apprehension about new technologies, there is also a pressing need to lead in global tech to counterbalance China and maintain American dominance. This represents a broader conflict over the future direction of technology and the innovation economy.
Northside Ventures
