5 Reasons Why Artificial Intelligence May Be Overrated

Is Artificial Intelligence Overrated? What if the Next Great Financial Crisis Is Accidentally Coming Our Way? Well, some people have asked. I recently came across this post by Greg Rosalskipodcast host Planet Money For NPR. I highly recommend you listen to it if the question that opens this article seems relevant to you. As a tech reporter who practically listens to, reads, or perhaps tries some new app supposedly based on artificial intelligence, I get the impression that this is indeed the case, and I pray that the latter never happens. In any case, after listening to your podcast and reading this article, I want to summarize the most striking aspects to me of why artificial intelligence is actually overrated.

It’s not really smart

Everyone is talking about ChatGPT, right? While it acts as a system that can practically answer the question you ask, this and other language models act as a kind of aggregator of the best that can be found on the Internet and throw it into a text box in a more or less coherent and readable way. If you ask basic questions like what is religion, or ask it to write an article about everything we know about the iPhone 15, the results will probably seem excellent. But if you ask it some kind of morally complex question that requires some fact-checking, it will probably end up saying something outrageous.

Your intelligence “source” is running low or is already charging you

Outside of the technology and economic environments, the case had little resonance, but several media companies, such as New York TimesThey have started suing companies like OpenAI for using their papers as “food” for their language model.

Following the deal, which was valued at about $250 million, News Corp., a conglomerate such as Wall Street Journal And New York Post— Allowed OpenAI to access your current and past publications. While $250 million may be considered a small amount in the tech space, it adds up to other costs, especially in the energy sector.

He makes a lot of mistakes.

And AI companies pompously call these errors “hallucinations.” Of course, AI companies aren’t particularly interested in revealing how “amazing” their models are, but one study puts the figure at between 3% and 27% of cases.

Mistakes may be tolerable, but the appeal of AI is that it will automate certain routine processes. The ATM Case Drive-thru cafe from McDonald’s, which made funny mistakes How to add $222 worth of nuggets to your order or suggesting to put bacon on ice cream is quite telling: it may solve some problems, but it creates problems that are very easy to avoid.

Still no killer app

Right now, the main applications of generative AI are in marketing and customer service chatbots, which, let’s be honest, no one really wants to talk to. With the exception of photo editing tools on your phone, you probably don’t use AI tools much or at all. I could use it to write a few articles here, but I’d have to spend a significant portion of my writing time reviewing its style and fact-checking, so I prefer to write.

It costs a lot of money and consumes a lot of energy.

We recently talked about the amount of electricity that AI applications require, and how building small nuclear power plants is being considered as one solution. Well, that energy cost will ultimately be an economic cost that will likely be passed on to users, and without a truly revolutionary application, no one will likely want to use AI.

Could this be a bubble that is about to burst?