When Alex Irpan of Google writes about compute as the way forward for AI, you wonder how much of this is AI pulling on compute vs. compute (and the investment into chips) pulling on AI.
The whole article is as much about economics as about AI, in fact it conflates the two 1/x
It starts with artificial general intelligence being equated with "economically valuable work":
"artificial general intelligence (AGI) [is] an AI system that matches or exceeds humans at almost all (95%+) economically valuable work
2/x
According to Irpan, economics determines also how AI will spread:
"We also don’t have to build AI systems that learn like humans do. If they’re capable of doing most human-level tasks, economics is going to do the rest, whether or not those systems are made in our own image."
3/x
here he exposes how the promise of AI begets more funds for compute, going full circle:
"If AGI is possible soon, how might that happen? [...] It would likely be based on scaling existing models. And, because it needs to be based on scaling, and scaling needs funding."
4/x
Here, the business model for tools: it is about capturing business budgets (by optimizing organizational workflows & productivity) while also selling expensive compute (hence the idea that more data could improve models a great business proposition for cloud providers) 5/x
Not all is new: the business model for tools has been how software companies have profited for years. Irpan mentions Lotus notes and MS Excel. @xiaochang does wonderful historical work of how IBM pushed for data intensive solutions to sell their hardware. 6/x
While computer scientists may throw around large words like intelligence, my favorite part of this article is how it shows that industry is ready to equate the pursuit of some definition of intelligence with profit. 7/x
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.