e.w. niedermeyer: Embrace Weizenbaum Thought — Subium • Bluesky web client

Embrace Weizenbaum Thought

MIT robotics pioneer Rodney Brooks thinks people are vastly overestimating generative AI

MIT robotics pioneer Rodney Brooks thinks people are vastly overestimating generative AItcrn.ch He says that the problem is that generative AI is not human or even human-like, and it’s flawed to try and assign human capabilities to it. © 2024 TechCrunch. All rights reserved. For personal use only.

If you want to understand AI and robotics, listen to the guy automating warehouses (process-driven drudgery in a controllable domain) and not the guys saying they are replacing creativity or creating digital gods.

Repetitive tasks coded in with bar codes, qr codes, where the machine follows precise instructions on movements. Sure. Replicating Mozart. Oh hell no.

Stares at bookmarked Rodney Brooks essay from 2017 ( rodneybrooks.com/forai-machin... )...nah essays and books and thinking are long let's go outside and assume bandwagons eventually collapse.

Even with the generative AI side, the best use cases I’ve seen are replacing drudgery rather than creativity. And in business there’s a lot of drudgery! It’s a huge market even if not glamourous

I dont understand why he wouldn't want his warehouse robots to have the personality of Brad Pitt.

Why does he hate science?

Guessing the Brad Pitt Mode robots may not be great for workplace productivity

No?

An interesting note: warehouses are very frequently not automated! We do, in fact, automate many. But it's a tool that works in some environments and not others instead of a one-size-fits-all needs approach. Not a matter of scale either.

Many warehouses use multiple approaches at once. They might have a robot placing and picking some items, workers grabbing others. Or they might use automation for aspects of the job but not all of them. I submit that even AI way more advanced than what we have will be the same.

Many warehouses find that they are allowed to pay people less to do the work than it would take to pay engineers to automate.

Generally speaking, there are AI developers who have credible use cases who are selling useful products and also there are charlatans looking for rubes to sell stories of magic. Keep the cow. Avoid the magic beans.

...or someone who can calculate power consumption...

Replace the Turing Test with a Weizenbaum Test.

Prove it MIT <<<

Just read Weizenbaum’s wiki. His decide/choice thoughts are profound. Thanks for the prompt.

Post