The Decaying Feedback Loop

Lets be real when you put Garbage In you get Garbage Out = GIGO.

This is the problem with AI.

A slight error repeated can lead you on the wrong course and the worst part is it happens so subtly you probably don’t even notice.

When over 50% of the content on the web is AI generated, then the majority of the training data to train new AI models is ….. you guessed it, generated by its previous version.

This is where the decaying feedback loop comes in.

What is it?

The decaying feedback loop is the function of generating an output and then feeding it back into the input.

With now over 50% of the content on the web being generated by AI, AI has to “learn” or “train” on a majority data set produced by itself.

Its pretty hard to get any information gain from that.

What does it mean?

This likely means that new LLM models will hit an asymptote and potentially reach a threshold which they can’t advance beyond.

This will be bad news for the hype and investment cycle that relies on a linear or even exponentially growing trend on capabilities and “intelligence”.

What do we do about it?

Critical thinking!

That’s it!

Those who ask questions and use critical thinking will do very well.

Those that blindly trust and copy paste AI output will not.