A discussion on the polarizing, hot-button issue of AI. Many laud its uses, others frame it as “useless”. The truth is never black and white. AI (and LLMs) are still very much a nascent technology, at least as I see it in the grand scheme of things to come. But, given the proliferation of this technology throughout the tech world, permeating much of our regular daily lives, you’d assume it was much more mature and robust. Below, I try to point out some of the many concerns that I have observed, in order to juxtapose those against the reality that AI/LLMs do in fact have real-world uses, and when employed correctly, can admittedly be powerful tools.

You can write something on this subject… but I’d scope it a bit differently, because their’s a lot more at play here than just “is it useful” vs “is it not”. To make a fair, leveled case for why LLMs have value, I think it’s worth first acknowledging all the things that are off-putting about LLMs. To name a few… (I’m going to use the terms “AI” and “LLMs” interchangeably throughout…)

• The seemingly devastating environmental impacts (and general resource consumption concerns)
• The whole “we had to rip off everyone’s writing, art, content, and general IP to make this work” aspect
• The plenty of use-cases where companies are trying to shove “AI” into things that just don’t make sense. (I saw someone post about an AI-enabled toilet… what?)
• The uncomfortable reality that LLMs (or generally “AI”) have been employed in place of actual humans, leading to workforce reductions, blah blah. (I understand this is the nature of technology advancements to some degree, but there’s still a human cost here that can’t be discounted, and just saying “git gud” to folks who are impacted isn’t really the best argument)
• The embarassing AI fails - e.g. remember when Gemini told us to put glue on our pizza?
• The scenarios where AI is used in ways that are actually not that great for AI
• The general griftiness and undeserved grandiose claims from the tech-bro CEOs that keep promising “huge leaps”, and “AGI is just around the corner”, but still can’t do seemingly basic shit a normal human can do…

Because look, I don’t think most people, if you were to ask them face-to-face and get an honest, tempered reply, would say that LLMs are literally “useless”. Instead, I think they would just point to any combination of the bullets above and try to make the argument that it’s just not worth the overall hype we’re seeing across the industry and the world at large. It’s just annoying. It’s jammed into every single product, and in these early stages of the tech, it’s had some really public face plants. Sure, it’s set to get better, tech has a way of doin’ that, but you have to admit the souring effects these fails have. So when we (the AI doubting populace) see these constant stream of visible fails coupled with surging product costs (to offset maintaining AI systems), mass layoffs (again because companies think AI can replicate those functions), and reports about how catastrophic this all is for the environment, we get to the point where we make statements like… it’s useless.

So, maybe without even diving into all those things you could just phrase your post as something like… “Yeah AI (or LLMs, whatever) suck for so many reasons, but they’re not COMPLETELY useless. Here’s a few things they’re OK at…” 😅