I get what you're saying, and you're right that a lot of early tech was janky. But it was also built to solve problems for the user, not to self destruct within 5 years so they buy another one. Engineers and experts in companies used to have more sway when they put their hand up, now there are professionally designed mechanisms in companies to silence and undermine them.

I grew up in the 1990s, and there genuinely was a shared belief that technology could make life better, that systems could become more humane, more efficient, and more empowering over time. Even when the solutions were rough, they were often built by people who cared deeply about understanding the problem and improving things incrementally. Not to boost their share price or go viral on social media.

What feels different now isn't that the tech is imperfect, but that large parts of the tech industry have deprioritised understanding and craftsmanship in favour of growth, optics, and financial engineering. The jank used to come from constraints; today it often comes from misaligned incentives. I am weary of attending 'tech' events because the focus isn't science, innovation or improving peoples lives - it's about fundraising and networking.

If you believe the hype, the returns should be astronomical. Even if they just deliver a fraction of the value promised they are huge game changers.

The core issue is that, with most things Ai, solving 90% of a problem is usually easier than solving the last 10%. You can't have autonomous cars that can't follow directions from firefighters and cops and you can't have Agents that will blow up your work flow when stumbling into something unexpected. When you have automated "all purpose" machine, those "edge cases" end up opening you to a lot of risk that can hit in an instant. That risk is what's really holding things back.

If the industry had paid attention to the autonomous vehicle landscape they'd seen that rolling out LLMs was going to be a long and arduous process, meaning slower and smaller returns than being promised.

The problem is that the vast majority of people don't understand the tech or what it's capable of.

And for once, this is kind of a real "both sides" argument.

The pro AI side is delusional and think LLMs can do anything, or will "eventually". You also have a bunch of fascists pushing the tech because generative AI is useful for propaganda in a number of ways and they want to use it to make shitty media because actual artists have generally been anti-fascist.

The anti AI side is more right, but they still buy into the bullshit about what the tech can actually do. They also think AI is going to take all the jobs even though it fundamentally can't. They ignore that while some companies might try to replace workers they quickly and quietly reverse course and rehire people.

And I'm basically convinced that a lot of the narrative of how the average anti-AI person talks about it is driven by the AI companies because it helps them sell their lies about what the tech is capable of.

Also, both sides ignore that most of the reports of layoffs because of AI are just companies that are using AI as a cover to not freak out investors, but they aren't actually replacing workers with AI, just reducing the number and overworking the rest.