Self-cannibalization of AI

It is highly interesting to see how certain topics are hyped in the age of think tanks and agency reports in order to make money. This can be observed in all areas, to the extent that weak people want to go to war with a cheer or pay seven times as much for the same product just because the packaging is green. In Germany, we also say “the same in green” when we want to sell people something they already have.

If you look at the major US venture capitalists, they clearly and honestly explain (it's about money) that #Blockchain, #Crypto and Eco are now over. Nobody believes it anymore. And now all forms of #AI, #AI and neural #networks are being invested in. But this #hype is also completely crazy, completely pointless and ultimately a paper tiger. Let's just hope that not so many children die this time.

The way neural networks are being unleashed on the disinterested public is completely superficial and pure marketing. After a year of hype, everyone recognizes an AI-generated image immediately. AI-generated texts are infantile and simply wrong in terms of content. You can't help but get the impression that it's all just hot air and that this technology, if it ever grows up at all, is currently in the stage of an unborn embryo that has been dragged out of the womb.

And like all premature babies, it will probably die. And that is primarily due to the training data that such a neural network needs. And they are cannibalizing each other. A good example of AI use is DeepL. They obviously use AI only as a tool and create wonderful natural translations from the writings and books of mankind.

But the hyped-up AIs, chatbots, image generators and music AIs have to resort to information from the uncensored internet. As everyone knows, this is a source that is not at all reliable, in which 99% of the content is made up and invented and (this will be the death knell for this type of AI) is now even generated by AI itself.

I call this self-cannibalization of AI. It is, in a sense, unstoppable. Neural networks feed themselves over and over again with their own data and structures. It's like incest in a rapidly reproducing population. It will degenerate and eventually die.

The large companies and conglomerates that are now betting on AI are in great danger of getting caught in a destructive downward spiral.

Or have you recently done a #Google search? And were you satisfied? Well then.

https://word.undead-network.de/2024/05/30/self-cannibalization-of-ai/