Do you think investors are going to just sink money into AI forever without a clear path to profitability? None of the big AI players are profitable - not a single one of them. Your free ChatGPT queries are costing OpenAI money. Even your ChatGPT Plus subscription still does not cover their costs.

Tools like Claude Code and Codex are also priced below what it costs to run them.

These investments are mostly based on the promises of AGI - the path to which remains elusive. How long will investors keep throwing trillions at this? I don't know, but it's clearly not forever. This is NOT sustainable.

Also we are seeing actual signs that build-out is slowing: https://thedeepdive.ca/openai-.... Micron is left holding the bag after OpenAI decides not to purchase nearly as much RAM as they said they were going to.

Lacking any sort of meaningful progress on AGI, it is inevitable that the investment is going to pull back and AI companies will be left holding the bag. Your Claude Code subscription will triple or quadruple in price to reflect its true cost.

TikTok is more oriented around interacting with a feed full of algorithmically-selected videos from various creators you don't know rather than being a social network where you seek out specific users and add them to your friends / circle and chat with them regularly. I get that there might be some people who use TikTok like the latter, but I'm doubting that's super common.

So regardless of any ulterior motives behind shunning E2EE, who does this really affect in the scheme of things?

Kids only have access to the Internet if someone gives them that access. Generally speaking, that's going to be at home, on a phone, at school, or on a friend's device.

Parents already have the tools to regulate what their kids can do online through existing parental control tools baked into mobile OSes and apps. So a solution like this is superfluous.

Schools lock down their devices already, so that's a non-issue.

If they're on a friend's device, that's not something age-verification legislation can address unless they expect the device to take a face scan every 30 seconds to verify that the user using the device is the one who authenticated their age already.

So again I ask - what problem are we solving exactly?

The author is a CEO of an AI company and an investor in others. I don't mean to poo-poo the actual legitimate advances being made in AI, but this isn't an unbiased piece. He's overstating the capabilities of LLMs in my opinion, but someone who has vested financial interests in the industry is of course going to say those things. The current AI industry is based primarily on the commercialization of LLMs.

I use the latest coding models in my own work and while they do some really awesome things, they are far from "its replacing software engineers". I find they struggle on large code bases and sometimes hallucinate. For me that's okay, because as a SWE I can recognize good code from bullshit and tweak its output but it gets frustrating when it wrongly answers a question about your code (how does X interact with Y, where is the code for that) and it puts you on a wild goose chase around your code base.

LLMs are hitting a brick wall in scaling to the point where the very best models require an incredible amount of resources to run ( = expensive). The resource consumption of LLMs are increasing at a rate greater than what advances in compute power and memory capcity are increasing at.

I'm with Yann LeCun on LLMs: https://www.youtube.com/watch?...

They're not required to retain it and might even promise to destroy it when they're done, but these validations are often done by third parties and it's difficult to ascertain that everyone involved is doing what they're supposed to be doing according to policy and that what they're sharing amongst themselves is being done in a secure manner. As someone who has been the victim of multiple companies who have failed to protect my personal data according to their stated policies, I'm still going to say "no" to all of this nonsense, regardless of what they promise.

Stores with self-checkouts are already using it to detect theft. I know for a fact that both Walmart and Home Depot employ this. It uses both cameras and data from the self-checkouts to find behaviors such as concealing items or deliberate mis-scanning to cheat the system.

Home Depot's AI system falsely labeled me as a shoplifter when I was checking out. I bought some simple plumbing fittings and made the "mistake" of scanning one item twice (because I had two) and putting the other item in the bag without scanning it individually. It thought the second item was something else that I was trying to steal.

As soon as I made it to the exit, klaxons went off, orange lights flashed, and the shopping cart even locked up so that I couldn't push it out the door.

A receipt and bag check confirmed that I paid for everything in the cart, a terse apology was issued, and I was on my way. At least he didn't accuse me of theft, just wanted to check the bag because the system told him there was a potential discrepancy.

There are stories like this at Walmart, but with poorly-trained "Karen" employees who think the system is infallible and immediately start accusing shoppers of theft in front of other customers which has resulted in hurt feelings and threats of law enforcement over nothing. It has resulted in nationwide "re-training" of self-checkout employees.

Now I just get side-eye from the self-checkout hall monitors who stare at me when I'm checking out, sometimes getting into my personal space to look at my items as I'm checking out, but no outright accusations of theft.

The whole experience doesn't feel particularly friendly and the system doesn't seem like it's doing anything other than making customers feel "watched" and employees on edge for shoplifting that is false alarms most of the time.

Facebook rebranded themselves "Meta" and dumped tens of billions into the "metaverse", a concept that few people are interested in and show no sign of ever being interested in.

Companies can and do make dumb decisions that don't pan out and double-down on those decisions all the time. Why is AI different to you? What does the investor class know that we don't know or can't read for ourselves?

The promise of AI is "AGI", an ill-defined concept that will supposedly replace hundreds of millions of expensive workers with relatively cheap computers that never tire out, don't demand raises, etc.

AI has made some great advances over the last few years, but it's nowhere near the point at replacing anyone except maybe the lowest-level customer service agents who are only allowed to read from scripts. AI has been used as an excuse for companies laying people off, but only to paper over the poor performance of the company, not actual AI replacement of workers.

AGI will come, eventually, but it likely won't be like the LLMs of today. Investors will likely grow impatient with AI companies and pump the breaks on investment long before that happens, and demand something more concrete before showering AI companies with billions again.

Jeff Geerling had a similar issue when he found that some of the advertised features of a Bosch dishwasher were locked behind an app and you had to connect it to WiFi to enable:

https://www.jeffgeerling.com/b...
https://www.youtube.com/watch?...

I think in his case he was so invested in time with installing the thing before he found out about that limitation that he just sucked it up and kept it. We all encouraged him to take it back to make a statement to Bosch and the retailer but I get why one would feel defeated at that point.

(For the record, I don't have a beef with smart TVs but I draw the line at WiFi kitchen appliances)

The people who host social media platforms are just as entitled to freedom of speech and freedom of association as you or I are. You are suggesting that a platform that doesn't want to associate with you (or any specific comments / uploads) should not enjoy the same rights as others that would host your comment. This is a legally dubious position that I suspect wouldn't hold up in court. You would have to make *all* platforms liable for user content regardless of whether or not they moderate other unrelated comments.

Section 230 is about civil liability for platforms that host user-supplied content. It's about people suing platforms because they think your speech caused some kind of harm to them and they ought to have done something about it before that alleged harm happened. Nothing to do with the government issuing fines.

The issue isn't that platform don't have the right to moderate as they see fit without 230. The issue is that platforms can become liable to all sorts of novel litigation if they don't heavily regulate every bit of content (comments, videos, etc.) that a user posts. They will have to have armies of lawyers on staff to constantly research all the ways they could get sued for something you might post and extensive resources to police the content more so than they already do.

230 by and large shields platforms from this and forces anyone who has a beef with your content to just go after you instead of the platform.

Not having 230 protections means platforms will likely heavily clamp down on what users can say, do, etc. to the point that you will likely see a lot of websites (mainly smaller ones) shutting down simply because the liability is too great and more draconian moderation from big tech websites that can afford to do moderation at scale.

CEOs are almost always the highest compensated in the company but at a large company like Walmart I guarantee you that the combined cost of all the employees far exceeds that of the employees. Not having a CEO (or not paying him, whatever) wouldn't really change the financial situation of a company.

Now a company where it might actually be the case that the CEO makes way more than all the employees put together is Elon Musk and Tesla's extremely ridiculous compensation package for him. His pay package is actually a large percentage of the company's budget.

That's already a law federally. CFAA.

Laugh all you want but we actually use this.

We tell our customers to "add +100GB to your virtual machine" before upgrading to the latest version of our software and they interpreted it as "add +100GB to the VM and then go in and start extending partitions" but they end up extending the wrong thing and causes headaches.

You can't reduce the root filesystem while it's mounted, so we give them this ISO and tell them exactly what to type to fix what they've done.

The 32-bit thing doesn't matter though. If you still have systems that aren't 64-bit capable then just keep using the old ISO.

It's software. Apple can just offer to help them reset their phones to get the malware off. Apple doesn't have to do any more than that under their warranty obligations. They're not going to lose millions of dollars because of idiots who have that happen to them.

At some point you have to make users responsible for their own actions. Apple can put warnings and confirmation dialogs up on the toggle that allows sideloading and force the user to confirm that they're opening their phone up to possible danger. If people still sideload malware after that warning, it's on them.