AI is overhyped, but it is still a mind-blowingly useful technology. AI can help you be more productive today, but tomorrow, it might take your job. AI can give people who can afford it an advantage and increase the technology equity gap for those who can’t. AI will be vital in helping humans find a solution to climate change, but it is part of the problem, too. AI is trained on the creative output of all humans and is taking work from those same creatives. AI kind of sucks, but I can’t live without it. Oh, have you heard of my subscription AI tool, Snowgoose? You should subscribe to it.
Overhyped…
Read enough articles, and listen to enough podcasts, and you’ll probably come to the same conclusion that I have. No one really knows anything about where AI is going—not even the people working on it. My personal opinion is that we are reaching a wall in the current transformer-based LLMs. Smarter people that I think that: Current AI models a ‘dead end’ for human-level intelligence, scientists agree Last year, I would have thought that AGI or ASI was just around the corner, but now I do not think that we’ll achieve it without a significant breakthrough. Now, of course, that could always happen. And it could happen quickly. We just don’t know. AI is expensive and it needs the money flowing in from investors to stay afloat. OpenAI, Anthropic, Google, and others need the hype to continue.
…But Still Impressive
I use AI daily. I am pretty certain of the types of responses I will get when I prompt it. But every so often it makes my jaw drop giving me something impressive, something that, at least, feels like some reasoning went into it. All of the knowledge of everything is at my fingertips and I can just ask. I’m a better coder than AI, and I’m a better critical thinker, but I don’t have all of that knowledge in my brain that I can recall at a moment’s notice. AI has helped me think better, code faster, and become smarter. I’ve leveled up skills in days that might usually take me months. If AI progresses ended today, AI would still be one of the best tools we can use to augment our cognitive functions.
The Two Faces of AI
I used Google Deep Research to help research this blog post and Snowgoose (via o3) to help outline it. It took a lot of effort not to take the next step and stick that outline into Claude 3.7 Sonnet and have it just write the blog post. But then, at what point am I in the equation? Full transparency: My previous blog post: # Building Snowgoose: A Technical Deep Dive into a Unified AI Frontend was almost entirely written by AI based off my extensive technical documentation of the project. Since that technical documentation was written by me, I felt like I was still in the loop on that one. This is the conundrum. When is AI helping and when does everything just become AI slop? What is the good use and what is the bad use? When does AI help improve us, and when does it actively harm us? My goal is to be thoughtful enough in my AI use to understand that difference, but there’s no guarantee that I will. The siren song of just letting AI do everything is very strong.
The Anxiety-Inducing Future
I mostly think AI right now is fine. It’s a great tool. We should all be using it. It does use a lot of energy, but maybe so do a lot of other things we use. Most of my trepidation is about the future. We are already seeing AI disrupt the software engineer market; it can’t replace an engineer yet, but it can increase productivity enough to need less of them. We are seeing the quick degradation of the internet as more and more AI-generated nonsense gets posted everywhere. OpenAI and Anthropic are charging hundreds of dollars a month for the use of their premium subscriptions. API use like what Snowgoose gives you helps give access to the top models for cheaper, but what happens when those companies start gatekeeping their best features behind their subscription plans (we’re already seeing it with Deep Research and image generation with OpenAI)? What happens when students who come from rich families can afford the best in AI tutors, but poorer students are stuck with lower-capability models? Heck, what about when OpenAI starts charging $20,000 a month for its best models? How will that affect wealth gaps?
So What?
So what? I don’t know. I’m going to keep using AI. Nothing I can do to stop the future. I just reserve the right to change my views on a day-by-day basis. I reserve the right to keep using AI while criticizing it. Anyway, you should subscribe to my AI tool, Snowgoose. It’s pretty cool and affordable and I use it every day like a hypocrite.
