Why LLMs will not replace humans

Pourquoi les LLM ne vont pas remplacer les humains - HYTRAPE

You may have heard that technologies like GPT and other advanced AI will soon steal our jobs. There is this fear, almost palpable, that intelligent machines will take everything from us. But let's be honest, that's not what's happening at all.

Today, even with all the goodwill in the world, these language models (LLMs) can only solve a tiny fraction of our technical problems. Come on, let's be generous, say 1%?😅

Generative AI is undeniably fantastic and incredibly useful. It can accomplish some impressive tasks, but it still can't solve the majority of the problems we had before it appeared. Yes, it opened up new possibilities and avenues, but most of the things that mattered before continue to matter just as much today.😜

Companies are spending billions of dollars developing machine learning solutions. And guess what? Much of this money is allocated to making predictions from tabular data and time series analysis. Two areas where LLMs are completely lacking.

Take a moment to think about this. To what extent is an LLM able to think critically about a problem, interview subject matter experts, frame the problem, collect relevant data, set up a labeling process, experiment, and start this process again and again?🤐

Answer: They can't.🙃

They can help us, give us ideas, facilitate certain tasks, but they will never do everything alone.

But of course, there is always someone who claims that the next version will turn water into wine. Seriously ?

If that's what you're thinking, you're not paying attention. We have reached a local maximum of their capacity. LLMs represent considerable progress compared to what we had before, but their potential for progression is short and we are almost at the end.🤓

So, if you think that an LLM will replace you, you are deluding yourself.

I'm not a soothsayer and I'm not going to try to predict what will happen next year, let alone a decade from now. However, some things are almost certain:

  1. Artificial intelligence will continue to improve. And a lot.
  2. Big language models won't take your work. We will soon reach the end of their race, unless we achieve accessible and democratized AGI (Artificial General Intelligence).
  3. People with technical skills will continue to thrive. More than ever.
  4. Many people will regret not having learned sooner. They will think there was no point in learning because of this fear, and will feel stupid.

So relax.

2024 is a great year, and you will still have a job at the end. In fact, you're likely to do even more, thanks to artificial intelligence. Maybe you'll have the chance to build something that was impossible a few months ago. But you will always hold the reins.

Learn to live with the times today.🏆