LB: if you are a founder of an actually useful business that uses "AI" technology in some defensible way, you should be glad that everyone viscerally hates it so much in public. the sooner we can get out of this "AI" hype cycle, the sooner your investors will stop demanding that you slather "AI" branding all over everything, the sooner you can talk about what your product specifically does and what value it provides and stop blathering about "AI" vaguely.

It's an implementation detail. You don't talk about your other implementation details like this. Imagine if every consumer app started their pitch decks with screenshots of github actions yaml. "we are democratizing github actions" / "imagine where the future of github actions will take us" / "society will be forever changed by our hosted testing environment"

if your product is a customer service PBX why would anyone give a shit whether you're using linear algebra or not. talk about results.

I think this is why GenAI sounds so much like a cryptocurrency scam to a lay audience. When Microsoft launched the Xbox they didn't spend hours talking about the wafer lithography they used for the GPU, they told you about video games. The iPhone announcement talked about listening to music and using a web browser, not the relative merits of capacitive and resistive touchscreens. But Blockchain and GenAI hype both have this obsession with minor details that really should not be user-visible.

@glyph And then, today, Figure 0 and OpenAI demonstrate a robot communicating with a human and performing tasks with relative dexterity in response to, but not directed by, a human.

If not for the GPT-esque pauses between input and response, this would completely blow my mind. As it is, I'm still astonished at the robot's (LLM's?) ability to portray an understanding of contexts.

@elight as an MIT media lab project this would be incredibly impressive. as a company, it's a tech demo for … what? a robot that can very slowly move objects around in an incredibly sterile lab environment, in response to verbal commands? Is this a cost-effective response to an actual problem? You are clearly meant to *infer* that this robot could autonomously do a factory or sanitation job, but that is not what we are actually being shown.
@glyph Indeed. I am maintaining some healthy skepticism as well. There is a lot that they didn't show us.
@elight I can't find it now but I remember a thread, I think on Twitter, by a roboticist getting annoyed about these types of demos. The entire rant was *just* about all the ways that dust can destroy expensive servos and how field maintenance is a nightmare, how it is even disruptive to manufacturing robots that perform a fixed repetitive task in a heavily controlled environment; from their description, autonomous robots that interact with the public are many years off _just_ from that problem
@elight and they finished with something like "and that is problem 1 of 47 that is not shown in these videos" (pretty sure it was in response to Boston Dynamics)