AI Roundup 010: Amazon's big AI announcement

Apr 14, 2023

After being conspicuously absent from AI news in the last few months, Amazon has entered the fray in a big way. It announced Bedrock, a cloud provider for pre-trained models.

Why it matters:

Much of the big tech competition has focused on consumer products (Bing, Bard, Copilot). Amazon is bringing the fight to the infrastructure level.
It's a little awkward for Replicate and Hugging Face, two existing ML cloud providers. But Amazon is aiming Bedrock at "enterprise-scale" AI apps, which may mean less overlap.
Amazon isn't afraid of picking fights on several fronts. Amazon's CodeWhisperer, a competitor to Github Copilot, is now free for all users.

Elsewhere in the FAANG free-for-all:

Meta open sourced Animated Drawings, AI that turns doodles into animations.
Twitter - which apparently doesn't have enough to do - is also moving forward with generative AI.
Should big tech slow down in its pursuit of the next big thing? We're now hearing from concerned employees inside Google and Microsoft.

AutoGPT was all over Twitter this week. It's a ChatGPT agent - an autonomous program that creates tasks and executes them in an infinite loop.

Why it matters:

ChatGPT, for all its capability, is "only" a conversation partner. Agents represent AI with organized thoughts, plans, and real-world impacts.
Stanford's groundbreaking paper simulated 25 agents within a small town - an 8-bit Westworld. They showed complex emergent behavior and were more believable than humans role-playing.
Of course, AutoGPT and its kin are basic demos. But their use cases are getting more advanced by the day. And I suspect we'll soon see companies put billions into developing agents.

Last week, Italy banned ChatGPT over privacy concerns. This week, more world governments are approaching potential AI regulation.

What to watch:

The European Data Protection Board (EDPB), Europe's privacy watchdog, created a ChatGPT task force. Meanwhile, Italy followed up with a to-do list for OpenAI to get ChatGPT unbanned.
The White House released an AI Bill of Rights - guidelines for building safe and inclusive AI systems. And the Commerce Department is asking for comment on AI "accountability measures."
China is also writing AI guidelines, so that AI systems "reflect the core values of socialism." Plus, the CCP is requiring security reviews for generative AI products before launch.