Generative AI vs Regular AI

Traditional "Regular" AI

Not all AI is bad.

Traditional AI has been used for years before ChatGPT and Genrative AI. It uses algorithmns and rules to complete a restricted set of tasks. Traditional AI is recommendation algorthimns, virtual game opponents, and even virtual assistants like Siri.1
Formulas and curated data sets power these algorithmns to make them highly specific and accurate. I emphasize this because generative AI DOES NOT DO THIS! Genertative AI frequently "hallucinates", which is when outputs that are not based on training data, are incorrectly decoded by the transformer or do not follow any identifiable pattern.2 One example of traditional AI beating generative AI is in math. The phone application PhotoMath uses traditional AI optical character reader technology to recognize written math equations, then solves the equation and explains how to solve the problem in steps. 3 The only time the program ran into issues was when either you wrote the problem wrong, or the equations needed to solve the problem weren't in PhotoMath's dataset.

ChatGPT and other generative AI is technically a statistical model. Meaning that it is making decisions based off of patterns seen in training data. Rather than solving the problem, it is using its training data to spit out what it thinks is most likely to be the answer. "For instance, given the multiplication problem 5,7897 x 1,2832, ChatGPT — having seen a lot of multiplication problems — will likely infer the product of a number ending in “7” and a number ending in “2” will end in “4.” But it’ll struggle with the middle part. ChatGPT gave me the answer 742,021,104; the correct one is 742,934,304."4

Generative AI

"Generative AI relies on machine learning to understand, predict, and create content from data. It takes a massive amount of data for generative AI to function, and machine learning provides the training that fuels the AI to produce its results."1
Generative AI does not truly understand the subject matter of watever it is creating, which is why it has such a hard time with math. These large data sets also encompass pretty much the entire internet, so it's datasets have information from trusted sources AND unreliable sources. The AI cannot distinguish jokes either, which resulted in a ridiculous rollout of Google's AI summary feature.5

Why Does it Matter?

It matters because a lot of people either don't know or refuse to acknoledge the difference. Most people who are "anti-AI" are only anti generative AI. This is because traditional AI is consistent, accurate, and relies on finite--and usually ethically obtained--datasets. Generative AI on the other hand, requires datasets so large that it is impossible for them to be gaathered ethically. It also hallucinates as mentioned before, and is just plainly inconsistent.
People who support generative AI will try to conflate the two to make it seem like generative AI is more important to our lives than it actually is. This is not helped by companies taking pre-existing features done by traditional algorthims as "AI features". Generative AI companies are so desperate for use cases that they are forcing themselves into traditional AI spaces.

While this tweet is a joke, it brings up a good question: why use generative AI to replace already functioning traditional AI? Music apps already have sophisticated recommendation algorithmns that work well. There is literally no reason to add in ChatGPT.