He isn't even an expert. He's a hype man, a salesman.

Frankly, he's engaging in a zero sum fallacy. Because it's not perfect, it can't be done. But it quite clearly has been done, successfully, with a drastic reduction in fatalities and injury. Is it perfect? No. But it's better than anything Waymo has been able to achieve with fancy intrusive useless sensors (there's nothing quite as infuriating as getting in a Waymo and having it just sit there confused about the homeless guy standing on the curb), but it's made by Waymo's biggest potential competitor - Tesla.

You forget that Tesla's now got a (small) working Tesla robotaxi fleet, and their capabilities are far superior to what Waymo offers.

That isn't half of it.

How many people in the audience own a Tesla?

He's talking to people who, if they are not personally intimate with fully self-driving capabilities in a Tesla, are at least aware of the capability.

It's like Steve Jobs going to Xerox to give a talk about why mice are superior input devices for navigation over just a keyboard.

a) and yet, nobody pays for anything in these situations, usually. Healthcare picks it up. The price is inflated specifically because of socialized services: what's about $10k in operational cost ends up with a bill for $50k because of collusion in the subsidized healthcare environment.
b) and? We need people to fail to be guidestones for the others.
c) then the fee would be significantly higher.
c1) "normal countries" do not have parks like in America.
c2) no, they are not.

These kids compounded stupid decisions. Instead of sitting and waiting and giving their water to someone else to go get help, they went ahead recklessly. It's also an exceptional situation. You don't design your entire system around the exception.

The sick thing is, this is probably the first time most of those kids have ever really been outdoors on their own. They did this, in all likelihood, without any parental involvement, oversight, or inform.

Anyone who's spent more than an a couple hours hiking outdoors would be able to tell you that an 18 mile, 6k elevation hike is a 2 day backpacking trip for people in good shape. lol

The artifact - the dashboard - is unimportant. He was just using an example that his intended audience would understand the value of: C-level executives who live by dashboards.

The process is what's important, and he wrapped it in Microsoft's proprietary ecosystem to bring value proposition of his platforms forward.

The process isn't new, it isn't specific or special to Microsoft, but it would resonate with his intended audience. It's a good process, and one which maps well onto other technologies.

I can trivially do these things today, and could not before:
* I can pull current market data into a local database and not only build new reports and analysis off of them but then feed them through complex workflows
* I can pull current job postings via API and holistically analyze them against my resume, with my intent and purpose in mind (and then I can circle back and trivially build a personalized reusable expert system based on those analysis artifacts).

All of this is of course, fully possible without AI. It's just not practicable: what would take me 6-12 months of concerted effort to make useful can be done in a couple weekend hours over the course of a month. Systems integration is hard, and time consuming, work, arguably more complicated in many ways than development of a singular platform because you've got multiple domains you've got to have expert understanding of to be successful.

No different than any other corporate spin about the inflexibility or uselessness of their platform. AWS does this quite well. "Undifferentiated heavy lifting" - while ignoring the absurd complexity and beta quality of their own services. Every issue I have with their shit, I experience a new and different bug or incorrectness in their documentation, and it takes years for it to get corrected. There are still bugs in Lambda that I discovered and reported 5+ years ago which haven't been resolved.

At every level. They must be using claude to make their business decisions, too.

It's not only being demonstrated, it's being reveled in. Astounding.

I don't know if this is just the symptom of the AI bubble, cronyism, or "late stage capitalism" (even though the resemblance to capitalism long since ceased). But it's a mess.

20 years ago, these guys would've been tried for criminality by now. Both Anthropic and OpenAI. They're making Enron look like amateurs.

I'm glad I am not the only one who found that film insufferable.

I'm glad the plague of 1990s "we're going to make a retard the hero" movies came to an end. Unfortunately, now we're plagued with them directing the films.

Congratulations, you just reimplemented REST with additional steps.

That's a model specific problem, to a large degree.

The "AI assumed something stupid and destructive" problem went away when I started using better planning skills and models which were not anthropic's.

I've not had any of those "wild destructive hallucination reasoning" problems with GLM or Qwen models, even small versions, which I fought with regularly with opus.

Sure, you can do that. You can do that with buggy code that compiles, or python, too.

For AI, you can also have another agent and/or model review and fix it. You can ask it adversarial questions until you get it right. You can ask it to explain the architecture and design (or: you can plan it out ahead of time so that you know the design and then verify it).

It's no different than starting on an existing code base... just faster.

"a bunch of AI bots doing the work of a whole team, is just a pipe dream."

You're not doing things right, then. Or very well. It's pretty low intensity to author/implement 5 substantial features a day, test, fix the bugs, run QA, and do 10-20 odd commits. Give or take, given the complexity of the plans/features and how much they let the LLM fill in the blanks and how thorough they were in scoping the implementation. Maybe 100M-200M tokens, depending on your model/skill/skills/etc. More, if it's a language like python or typescript (noisy, lots of bugs).

That kind of thing used to take a team a day. A good dev might close 2-3 bugs a day if they already have a thorough RCA done by someone else (which, itself, usually takes a 4-8 hours, depending on the issue).

Wait... you mean there were actually people trying to use as many tokens as possible?

It's one thing to try to exhaust a usage quota (which has never been difficult) because it's a subsidized commodity. It's another thing entirely to just burn tokens.

How can a CEO say stupid shit like this? Oh right. CEO brain.

There is no part of this which even aligns partially with reality.

Let's assume there's a catastrophic blow up with the AI market. Anthropic and OpenAI are gone, or crippled. Resold for parts. NVIDIA closes up shop or goes back to making video game cards after a restructure.

What then?

There's still incredible demand for inference. Every business is scrambling to get more, cheaper, better inference. Almost all of them - whether they're successful with it or just getting in the ropes. That demand isn't just going to disappear, and people/companies able to leverage AI will continue to do so.

People are currently still running - and buying - 2018-era GPUs for inference locally. In fact, most cards cost markedly more (and continue to cost more) than they did new. They're also more capable at running inference than they were even a short while ago: you can do things on a 12GB 3060 today which weren't even possible on the 24GB 3090 6 months ago ago. For the time being, these cards are still increasing in value.

Short of electrical failure, the forefront GPUs will be capable for both inference and training for a long time to come. "Depreciating assets" aside, the actual value will continue to increase short of systems with more capabilities being released.

Which, if the AI bubble pops hard, is likely to slow dramatically.