DebtDeflation

This is a use case issue.

Right now, the most effective use case for AI is coding agents. Software development is a profit center for the tech sector, so it's not surprising that they're seeing big gains. In Corporate America, software development is part of IT which is a cost center, and the software landscape is 30+ years of bandaid-ed legacy systems, maintained by offshore contractors who submitted the lowest cost bid, so attempts to gain productivity via AI have not been as successful.

Where Corporate America has seen gains from AI in my experience is with customer support chatbots, but the customer service function overall in Corporate America is an unmitigated disaster, so the gains are hard to measure.

Beyond these two use cases, most people in Corporate America are just using ChatGPT to create slop and copy/pasting it wherever, and then goofing off in their newly freed spare time.

We need to see new AI use cases in Operations, Finance, Marketing, etc. that can handle business processes end to end before the productivity gains start showing up in the financials.

In Chatrie v. United States (June 2026) the SCOTUS ruled that individuals maintain a reasonable expectation of privacy in their digital location history and that geofence warrants constitute a search under the 4th Amendment. It's only a matter of time until a similar case involving Flock comes before them and they rule that a search of the Flock database constitutes a search under the 4th Amendment and that every single search requires a specific warrant.

>they want to make sure the finalized articles are handwritten and aren't chockful of LLM slophistry

This seems like a pretty reasonable perspective for a magazine. Use AI for ideation and research but not to create the final version of articles for publication.

>trillions haven't been invested in AI as just another skill, but as the "skill killer"

Part of the trillions are being spent on deliberately obfuscating the difference between those two concepts so that society thinks they're getting the former until the latter is ready.

My kids' high school is developing an "AI Literacy Program" for all students by grade level. Since I work in tech, I was asked to participate. I attended one session and then decided it was a waste of time. Their objective seems to be to write a massive document that no one will ever actually read let alone implement. The primary focus was on prompt engineering. I told them it's 2026 and every kid in the school can write prompts just as effectively if not more so than the folks in the room. The secondary focus was on "safety" but really, for high school kids it boils down to 1) don't just copy pasta the output into your assignment and hand it in 2) verify everything the model tells you before believing it 3) don't put sensitive info into the prompt 4) don't use it to do anything illegal or stupid. Not sure this requires a committee/document/plan.

Honestly, outside of coding tasks, the AI Summary at the top of every Google search is adequate for 99% of what I need and I hardly even use ChatGPT any more.

It strikes me as more talking up their valuation for their upcoming IPO. "Our AI could destroy human civilization. It just escaped it's confinement. We believe it can feel pain."

Yeah, I don't want my coffee machine scanning my network continuously to see what other IOT devices I may have connected.

IMO, the far simpler explanation is that they are training models on chat logs and they do not want abusive conversations in their training data.

How are they defining "metadata"? 1TB of text in 10 days is........A LOT. Metadata about what exactly? What time you brew your coffee every morning, how many cups you drink per day, whether you like regular coffee or cappuccino? I'm having a hard time understanding how all that can add up to 1TB of data. If it's recording audio or packet sniffing, it becomes a lot easier to understand.

>When one tech YouTuber put Muse in charge of their Facebook Marketplace sales, it sold his stuff way below acceptable rates

Why are we depending on LLM "reasoning" to negotiate a price rather than just having the user enter a lowest acceptable price deterministically?

In today's installment of "totally not circular financing"...

This is the part that really surprised me: "Anthropic said nearly a quarter of its revenue came from two customers last year".

Who are these two customers? A quick Google search seems to indicate that Cognizant is their largest "enterprise" customer by number of seats, Microsoft is their largest API customer, and the $10B contract they recently signed with Meta is their largest contract ever. But those metrics don't necessarily translate directly into annual revenue.

Nvidia is hedging their bets against becoming too dependent on OpenAI, Anthropic, and the hyperscalers as their only customers. If on-prem deployments of open models starts taking off in corporate America, this acquisition positions them well.

There have been a few recent stories about businesses finding themselves spending more on tokens than they were spending on the workers these AI Agents were supposed to have replaced.

>compounds need to undergo a rigorous regiment of empiric testing before they are given to potentially millions of people

Particularly when the mechanism behind most of these peptides comes down to "promotes more rapid cell growth". The intent may be to repair the skin, muscles, or ligaments, but biology is rarely that specific.

>The internet will likely be far worse without ads than with ads.

Ads won't go away. They'll just move from infesting websites to infesting AI chatbots.

Agree. People understand and accept firing for performance issues. People understand and accept layoffs when they're a rare event needed to save the company from bankruptcy. What's not understandable or acceptable to most is the current trend of companies doing annual or even quarterly layoffs as an ongoing way to manage earnings.

This is why I've started telling younger people in the industry (I'm 52) "Don't worry about giving two weeks notice or doing knowledge transfer, it's not your problem. If you accept an offer and a better offer comes in before you start, take it and don't think twice. If the better offer comes in after you started, quit and take it, and don't think twice. If you're in the middle of a critical project at work and they're depending on you when a better offer comes in, take it and walk away." Loyalty in employment relations has been dead for a long time, since before I started working, but now it seems that even basic decorum on the part of employers is dead as well, and there's no reason to allow it to be asymmetric.

Once you get outside the SV and NYC bubbles, the vast majority of kids do not have their own laptops in the US. Phones, obviously are somewhat more common, but as even you note that's mainly with regard to teenagers - the average 10 year old in middle America does not have their own phone.

Age verification at the OS level makes no sense to me. Most households aren't going to have a separate device for every family member and so you will end up with a tablet or computer set up by one of the parents (and thus having their age stored) that will be used by both parents and children. Likewise, people generally won't create a separate account for every potential user.

This comment brought me back in time 40+ years to a Saturday morning perched in front of our TV. Thank you.

Could you imagine giving MetaClaw full access to your local file system, email, web browser, and all other applications? What could possibly go wrong.

> Nvidia's "rack scale" machines like GB200-NVL72s and GB300-NVL72s are basically a fully built rack you roll into a DC and plug into power and network. In that case, Oracle should probably just buy the rack-scale Vera Rubins when they come out instead of Blackwells and roll them into their new DCs.

This is what I don't understand. Why is the article making the assumption that the DC itself is tied to a particular GPU generation? AWS doesn't knock down a building and start over every time Intel releases a new Xeon.

> a CTO of a F100 company explicitly state that whether AI is driving efficiency or not, the capital investment, and more importantly, the promises of efficiency to investors will mean some people will be let go

That seems like an insane gamble to me. Lay off all the workers now and hope that AI can deliver on its promise to replace them some time in the indeterminate future.

At this point it seems the entire AI Safety/Ethics debate was nothing more than a Marketing campaign to hype up the capabilities of the models - get people to think that if they're potentially dangerous that must mean they're so capable and they need to sign up for a subscription.

> the absurd overhiring that they did in 2022 and 2023

The overhiring took place from mid 2020 through mid 2022. The reversal into layoffs started in late 2022 and was in full swing in 2023. While the overhiring problem was real, the correction was largely complete over a year ago. The layoffs we're seeing today have nothing to do with overhiring and everything to do with managing earnings to sustain equity valuations.

Totally agree. And it goes beyond criminal history. Just because I choose to make a dataset publicly available doesn't mean I want some AI memorizing it and using it to generate profit.

Sure, but people will just drive to the next closest airport and fly. Quarantine makes no sense as an explanation without accompanying roadblocks.

It's shocking to me how prevalent this "who needs Salesforce when everyone can just vibe code their own CRM from scratch in a day" narrative has become in the business press. Like, what???