Signs of a market bubble

So Mr. Zuckerberg personally met with Mr. Deitke. Then Meta returned with a revised offer of around $250 million over four years, with potentially up to $100 million of that to be paid in the first year, the people said.

This is completely wild. I don’t know how to interpret it. BTW, Deitke is 24 years old.

3 Likes

I can’t stress this book enough to those reading this thread:

2 Likes

The existing paradigm for AI is very new. The way it is trained, the products that are put out, the way humans collaborate with AI could all dramatically change.

The book is discussed in this video:

Silicon Valley Insider EXPOSES Cult-Like AI Companies | Aaron Bastani Meets Karen Hao

The video has some interesting points about the choices being made:

OpenAI started as a non-profit because that gave it an advantage in recruiting talent. After about a year of development, the bottleneck for OpenAI was finding capital, and so it switched to a for profit model.

Sam Altman is very good at inspiring people and convincing people, but is controversial depending on if you agree with his goals.

AI companies are searching for locations with water and energy. OpenAI focused on solving the AI problem through massive scaling, and this method consumes massive amounts of energy and water. There are other methods that could be used (tiny AI, DeepSeek).

There is a struggle between Silicon Valley and Washington DC, and neither sees democracy as important going forward. Both think democracy has failed in the US. DC has enabled Silicon Valley to place its technology across the world (the OBBBA gives tax subsidies to hyperscalers through accelerated depreciation), with the idea that DC will nationalize later and so build a future DC empire. Silicone Valley sees the way to organize people in the US as a network of companies. Imperial Conquest.

The Senate dropped the ban on AI regulation that was in the House version of the OBBBA.

New technology is only worth doing when it has a monopoly. Large AI companies are continuing to use the scaling paradigm because scaling produces a barrier to entry for their competitors.

=== link ===

The electricity demands of data centers are making it harder to build new homes in London, July 28, 2022
“According to a report from The Financial Times, which shared letters sent to housing developers by London’s local governing body (the Greater London Authority, or GLA), the issue is affecting new developments in three west London boroughs: Ealing, Hillingdon, and Hounslow. The GLA said west London’s electricity grid was at capacity, and told one developer there may not be “sufficient electrical capacity for a new connection” until 2035.”

3 Likes

@bjurasz

The book discussion on the linked sales page “breakneck pace”? Not according to one recent review of CGPT 3.0 and 4.0. Very little progress and financial losses.

Progress is stalling out.

It would take an internet 4 times its current size to train ai properly. Supposedly.