If you’ve been reading my ongoing series on why AI is so terrible for web development, you might come away thinking I hate AI.
And you’d be right!
Because AI isn’t real. There’s absolutely nothing “intelligent” about these tools. They’re the modern equivalent of a person in a robot suit pretending to be a fully sentient creature.
Nearly everything with an AI label on it is just marketing hype.
Even the stuff that’s been hyped as where these tools can actually be helpful is, well… a lie. For example, there’s been a lot of talk about using AI to help doctors catch diagnoses they might miss in the data.
What happens in reality, though? Doctors actually started missing diagnosis when AI provided (predictably) provided incorrect results!
When the AI provided an incorrect result, researchers found inexperienced and moderately experienced radiologists dropped their cancer-detecting accuracy from around 80% to about 22%. Very experienced radiologists’ accuracy dropped from nearly 80% to 45%.
But LLMs? They actually do have some uses!
For example, I think OCR (optical character recognition) has a lot of really positive applications. It can make digitizing massive amounts of archival text a lot easier, and can improve accessibility in the physical world for people who have vision impairments.
Similarly, I think the use of LLM-like tools to catalog giant digital photo collections by people, places, and things is a huge boon for a lot of folks. Being able to search for a photo that contains “dog beach” and easily find that photo of your childhood dog from that family vacation long ago is amazing.
But that’s not “AI.”
This isn’t new technology. Those tools have been around for years. They’re not “thinking” or answering questions. They’re pattern matching. Which is what everything people call AI is doing, too.
But unlike stuff labeled “AI,” they’re fucking honest about it.