Hate the Company Not AI - Short Rant

Let’s start this with a fact: AI has been generally very useful for us.

No I’m not talking about the generative AI hype driven by OpenAI and others but way before this craze happened in the first place. There are in fact a lot of things you don’t really notice and taken for granted which in fact, has been made possible by the advent of AI.

Take a classic example, spam filter. Have you ever got an email from a Nigerian prince in your inbox over the past few years? I have not, and that in fact is a classic problem that earlier AI models were able to solve. Your typo autocorrect? AI. Your youtube/spotify/etc recommendation? Also AI. Even your search engine, website protection such as Cloudflare all use AI to some extent, before the generative AI hype took place.

These are all, I hope we can agree on this, very useful technologies made for real problems in the real world. They are made to solve an issue that existed and has been plaguing our daily lives - and they solved it brilliantly. They are reliable, robust, and doesn’t cost a fortune to run them.

I think you can see where I’m going with this. Is generative AI useful? Yes, without a doubt. I think it’s fantastic as an information retrieval tool - it can help you gain understanding over a huge amount of data faster, enabling us to do research way faster and easier than ever. But what about the hallucination? I hear you saying - I think that’s fair. It’s definitely a drawback that we needed to solve, but the core function still worth something - it’s genuinely useful and we can see how we might want to invest further in it.

On the other side of the coin though, now the AI companies are increasingly being desperate trying to find use cases that can keep the hype alive and blow the minds of the public. We went from text generation, image generation, and now video generation with scary level of precision. And these are very expensive model that would not really contribute towards anything significant.

So what if you can generate images easier? Is there way more demand for images compared to artists that could draw them? Is there a surplus of demand for videos compared to millions of content creator that uploads their videos on daily basis? Is it really worth the billion dollars worth of investments to flood our internet with even more sloppily generated stuffs? Not to mention the risks of proliferating our internet with false information, scams, bots, and other generated stuffs which are close to indistinguishable without significant effort to dig deeper.

I really can’t think of a use case that could justify these things. The idea that we could fly to the moon with a computer thousands of times less powerful than our smartphone, yet now we require thousands of times more powerful machines than our smartphone to generate a 30s slop tiktok blew my mind away. In my opinion, these features are just distractions. Distractions that at the moment - AI companies can’t generate enough revenue/profit to justify their exorbitant spending. At the moment, they are able to keep up the hype and bubble with increasingly more reckless features - but there will be a point where they don’t and that’s when everything will crash to a halt.

TLDR: I don’t hate AI - they are useful - but I hate the direction of AI companies are heading towards. Solution should be made for a problem - not the other way around.