3 critical flaws that explain why AI will never live up to the hype

Generative AI is being shoved in our faces, and “AI” of the last two years differs greatly from “AI” of the time before it. Pre-OpenAI, AI was a buzzword for the most part that still carried meaning and still powered features we cared about. Your phone had AI that it could use for HDR photos, your laptop used basic AI to schedule tasks across processors, and your online algorithms like TikTok and YouTube used AI to determine what content to show you.

Since then, AI has become a whole different beast, driven primarily by companies that believe it’s the next best thing. While there’s some benefit to the AI advances that are happening today, for the most part, people are sick and tired of hearing about AI over and over again. Companies feel like they’re creating features first and looking for uses after, but why is that? What are the flaws that hold back AI from really permeating the market?

3 Copyright issues and public perception

Everyone hates AI before they even know what it is

The biggest problem facing AI right now is public perception, and that’s in large part thanks to its perceived copyright violations. AI learns from content on the internet, which many argue is a violation of copyright and is tantamount to stealing. After all, artists didn’t give permission for their art to be used to train an AI model, but some companies are even using AI art instead of paying artists to do image work for them.

Whether or not it’s legally a problem is irrelevant, as the court of public opinion has largely made its decision, which in turn leads to increased annoyance as generative AI continues to proliferate. It doesn’t help that “AI” has been conflated with “generative AI”, meaning that it’s all one and the same in the minds of the public. What once was mildly humourous to see plastered all over new products has now become a mark of the devil.

Even if you don’t care much about AI or know much about it, the copyright issues surrounding it are starting to get louder and louder. OpenAI even said it can’t train its models without copyrighted material, which in itself isn’t a great thing to say when your biggest haters really dislike copyright infringement of their own works.

Related

Best AI image generators: DALL-E 3, Stable Diffusion, and more

If you want to try out some image generators, these are the best ones on the market today.

2 What can you even use it for?

The use cases are few and far between

For most people, there are very few real uses for generative AI and the powerful models behind the likes of Copilot and Gemini. Sure, it can help summarize a document, but you still need to read the document after. It can create text for you, but you still need to check the text for you. It can even create images, but make sure people have the right number of fingers.

Generative AI is an incredibly powerful piece of tech, but for most people, it’s not worth even using. There are very few reasons to use it over the tools we already have on our phones, and the features that it offers, because of hallucination risks, aren’t even trustworthy. Even if someone thinks it’s useful, it only takes one flub where it misses something important in an email summarization for a person to never want to use it again. There are some great productivity tools out there that are powered by AI, but they’re few and far between.

1 It’s not cheap

Most services are paid

One of the biggest barriers to entry for generative AI is the cost. Pretty much all of the main services are a subscription or require some kind of payment in general. The compute power required is immense, and the costs associated with that are incredibly high. With all of the other problems surrounding AI, why would you pay money for it? OpenAI is burning through money at a rate never seen before in tech, with the company allegedly spending up to 7 billion dollars on training and inference and another 1.5 billion on staff for this year.

That’s just ChatGPT as well, there are other companies in the mix like Google, Microsoft, and Amazon that are all burning through piles of money. When does someone pull the plug when they realize it’s never going to be cash flow positive? That perception alone is also doing damage, because these services could disappear practically overnight in the future once they’re deemed unprofitable. It’s simply not sustainable, and no company can lose billions of dollars year after year and rely on investor funding to keep afloat, certainly not forever.