My new stance on AI, after over a year

Yuck. Disgusting.

It’s been a while since I’ve talked about AI, and a lot has changed on my perspective of it. There are definitely important distinctions to be made when talking about it, though.

Generative AI

This is the part that I’m grossed out about. Generative AI, by definition, is an artificial intelligence trained off of a ton of a specific form of content that attempts to recreate content based off of its training data. There are many, many reasons why it is bad.

For example, how do you think the groups that create and operate these models collect the data they use to train the models, with billions if not trillions of unique works collected all in usually less than a year or two? Well, I’ll tell you. It has nothing to do with legality. If it is on the internet, their bots will scrape it (unless you put a robots.txt file at the root of your server that explicitly bans bots, but not all groups honor this). It then gets used, and you don’t get anything from it. This also means that some chatting apps like Whatsapp will just go and collect all of your private chats and use them to train their AI. Bye bye, privacy!

Another great example is resource usage. Generative AI generally requires tons of computation power to work properly, so the companies and organizations running the AIs will build datacenters everywhere possible. End result? You end up with a massive datacenter in what is usually a small town that strains local power grids, leading to cost increases for normal citizens that aren’t participating in the whole AI business. It also needs water to cool down the servers inside said datacenter, which pollutes the local water supply and if released back into nature can end up killing tons of animals, aquatic life, and plants. Also, people drink that water, so it is a net loss. And I may hear you say: “But datacenters all over the world use water to cool down their servers, what’s the difference?” Well, here, I’ll tell you with an analogy. If there was a water reservoir, and a person took a glass of water from it daily, would you see it as problematic? No. But if 5,000 people were lined up to get a glass of water from the reservoir daily, then would it be problematic? Yes, especially once you take into account that there are other people that genuinely do need that water to survive. AI datacenters are springing up like wildfire, and so far, nobody has done anything about it, leading to major consequences like the ones I just described.

Also, what happens to creatives? They lose their jobs. People realize that instead of contracting professionals that spent so long mastering their own unique artstyle, or writing style, or music style, or whatever it may be, they can just go to an AI and have it pump out slop after about a minute or so of generating files. Soulless slop that misses the whole point of the arts itself, which is to bring connection with others, joy, and also just act as a creative outlet for those who wish to make something. Much of the economy depends on artists, musicians, writers, programmers, and similar titles in some sort of fashion, and we will see the consequences of abandoning them just to save the extra buck.

Non-Generative AI

I don’t mind AI that isn’t generative, and it has been around for years at this point.

Non-generative AI does have plenty of uses, and there are tons of them that we use all the time, and don’t even notice. For example, translation. Many modern language translation engines do use AI, or more technically, neural networks. They are trained upon tons of different proper translations, which then allow it to properly translate practically any language if it is given enough examples of proper translation from said language to any other.

Another great, positive example of non-generative AI can be seen in the medical field. Often, a properly trained neural network will be given an x-ray, or medical research results, or something similar and it will be able to classify what the issue is: such as, (and these are really basic and oversimplified) “No issue”, “Respiratory”, “Cardiovascular”, “Muscular”, etc. They have existed for decades at this point, constantly helping doctors, nurses, and other sorts of trained medical professionals.

This one is a little more obvious, but AI is in a ton of games. For example, in Mario Kart, there is a neural network (or a similar variant, at least) constantly running that decides when your opponents are going to throw an item, turn, drift, or attempt to slipstream off of you (get a boost by being behind you). Or, in a FPS, like Counter Strike, the bots that are fighting against you or with you are being powered by an AI that controls every single action they make. Many games usually offer an easy, normal, and hard mode for the AI, and that typically just means the AI getting more or less training, or it being trained off of better or worse players.

Conclusion

Anyway, that’s all I have for now. If you want more tech tips and digital blabbering, save yourself, the economy, nature, and most importantly, a bookmark of the CheeseBlog to to stay tuned for more posts. Cya!