The AI dooooooom thread

[The Minsky Moment]

The two possible states of AI:
1) AI is too stupid to be really useful (where we are now).
2) AI is clever enough to be useful but then humans are too stupid to safely control it (where we are going to)

Either way we are out a few trillion.

[crazy canuck]

There is a third option, AI tools produce work that is mediocre and humans put too much trust in it's capabilities and are not putting enough effort into verifying what the AI tools are doing.

Also ends in disaster

[Jacob]

Iranian strikes test the Gulf's trillion-dollar AI dream
The region sold itself as a safe harbor for the world's data. Amazon's burning data center in the UAE has upended that pitch.

For years, Gulf leaders made a simple promise to Silicon Valley: Bring your data, your models, and your chips, and we will give you stability.

On Sunday, that promise ended in flames, after Iran's retaliatory strikes in response to U.S.-Israeli assault set an Amazon data center in the United Arab Emirates on fire.

In a statement released following the incident, Amazon said "objects" struck the building, creating sparks and flames, declining to link the incident to Iran's missile and drone attacks. Fire crews cut power to the entire site, and more than 24 hours later the facility remains offline, with the disruption spreading to other parts of Amazon's UAE operation.

Full article here: https://restofworld.org/2026/amazon-uae-data-center-fire-iran-strike/

Crossposted to the Iran War thread

[Sheilbh]

The two possible states of AI:
1) AI is too stupid to be really useful (where we are now).
2) AI is clever enough to be useful but then humans are too stupid to safely control it (where we are going to)

Either way we are out a few trillion.

I think it is one of those things where the future is here, it's just not evenly distributed.

[HVC]

Microsoft banned the use of the word Microslop on its AI discord channel . Which brings up the question of why Microsoft has a discord channel.

[Valmy]

Iranian strikes test the Gulf's trillion-dollar AI dream
The region sold itself as a safe harbor for the world's data. Amazon's burning data center in the UAE has upended that pitch.

For years, Gulf leaders made a simple promise to Silicon Valley: Bring your data, your models, and your chips, and we will give you stability.

On Sunday, that promise ended in flames, after Iran's retaliatory strikes in response to U.S.-Israeli assault set an Amazon data center in the United Arab Emirates on fire.

In a statement released following the incident, Amazon said "objects" struck the building, creating sparks and flames, declining to link the incident to Iran's missile and drone attacks. Fire crews cut power to the entire site, and more than 24 hours later the facility remains offline, with the disruption spreading to other parts of Amazon's UAE operation.

Full article here: https://restofworld.org/2026/amazon-uae-data-center-fire-iran-strike/

Crossposted to the Iran War thread

Gosh Jeff, maybe you should have let the Washington Post endorse Kamala.

I hope his fucking trash data center burns to the ground.

[The Minsky Moment]

Amazon said "objects" struck the building, creating sparks and flames, declining to link the incident to Iran's missile and drone attacks.

Is Kristi Noem Gaslight Syndrome contagious?

[Syt]

From Palantir's Youtube:

Cameron Stanley, Chief Digital and Artificial Intelligence Officer of the Department of War, shares how the DoW is driving enterprise-wide adoption of data, analytics, and AI to generate decision advantage—and what it takes to move cutting-edge technology from the lab to the warfighter at speed.

The use of "enterprise-wide" for the Department of Defense WAAAGH seems weird but fitting with this administration. (Though I guess corporate influence on defense purchasing has always been strong )

[Syt]

Tbh, the gamer in me thinks, "Why can't we get this kind of interface for games like Command: Modern Operations?" But I guess the devs would be accused of "dumbing it down" and that "it looks like a mobile game."

[bogh]

A very real impact of AI will be the extreme drop in the cost of code. Developers are becoming incredibly productive and AI eliminates many of the drawbacks that previously haunted building inhouse over buying software as a service (documentation was hard to maintain, taking over someones code was problematic, best practices were only valid as long as everybody followed them etc.).

So while SaaS companies are currently stuffing AI into their products and jacking up prices, it will be interesting to see the longer term effects here. One could easily imagine a market where companies increasingly revert to bespoke and custom solutions, undermining the traditional SaaS model, but potentially increasing productivity as systems support their specific needs better. So there is probably a boost from AI that's not about AI itself, but from the effects of code dropping massively in price.

[HVC]

Maybe for a bit until people start retiring and you don't have anyone to replace them because haven't trained any junior employees because they've all been replaced by AI.

[bogh]

I am not talking about replacing anybody with AI, simply the fact that the cost of building specific tools (that may or may not have AI functionality in them) is coming way down. Even in a sector with zero AI implementation will be impacted by the fact that the non-AI systems they rely on can be built much cheaper, much faster and much more customised for their specific business.

[Zanza]

I saw some Antrophic multi-agent setups with the latest Claude model in two use cases last week (coding software, evaluating supplier offers) and it was impressive and scary. The agents can really coordinate well now and understood the context very well.

My company is starting to count AI agents in our FTE counts for software development now. If you book a team of programmers the AI support is factored in so that you can either get more output than before or pay less. Biggest factor is in legacy code migration.

[bogh]

Yeah. In (pro) code the hype is definitely real. Very impressive and a genuine revolution.

[Crazy_Ivan80]

Claude is insanely good. Especially compared to 6 months ago when we still had to babysit it constantly.

The result is that we've been burning through our backlog as well as quarterly road map. Which is both good and bad. Good in that we can do more, bad in that our qa can't keep up.