A big fail of Meta’s newest program is a case of AI failure that happened in recent times.In fact the public demo was available only for three days before it was taken down due to a huge wave of critisism and errors that could result in a lot of trouble…
But first of all what was Galacitca ? The AI model developed by Meta was meant to be a great tool for scientific research. The program was supposed to be doing all the job for us. Gathering information , articles or even things like : solving equations and creating statements based on the combined knowledge. It sounds like a perfect device , but the results were rather unsatisfying.
The basic problem that has begun to develop was that the statements created by AI were bizzare and incorrect. Program started to build articles that contained informations that seemed to be reliable, but there was no actual knowledge inclued. The case was very serious, because some of the articles reffered to real people, but the sources were fictional.
Another inaccuracies appeared in filtering systems , where subjects like for example racism or AIDS disease were unavailable. Michael Black the director of the Max Planck Institute posted his concerns about this project on twitter platform stating : “It is potentially distorting and dangerous for science.”
Right after i read about galactica i had two things in my mind : great admiration for AI ( how fast it evolves etc. ) and great fear of how this technology might be used to cause damage and general disinformation. unfortunately I had a strong association with “rewriting history” phenomenon that has occurred for example in soviet union. I think that such programs should be tested much more precisely before being published , because a lot of people these days cant verify the informations that they have been given. Which can as I already said cause some trouble or be used for malicious puropses by hackers , terrorists or other evil ones…
So as not to end on such an unpleasant note .I personally think that general idea of Galactica was brilliant. A great programme that could be a huge time saver, however, we can see that it will still be a long time before we can use this software freely. We need to be aware that it is still a bot that is not (yet 🙂 ) able to filter the real fact from the fictional one.
I am curious to see how this model will develop and how quickly the team responsible for the project will get over these problems. If they manage to overcome the obstacles at all.
Tell me what you think in the comments ↓
sources :
https://twitter.com/Michael_J_Black?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Eauthor
https://pbs.twimg.com/media/Fhrc_3TWAAE-u8G?format=jpg&name=medium
Actually, I couldn’t add a comment to your fashion digitalization blog so I will write it down here. So I totally agree with you I love customising m characters in games it really gives you satisfaction. Especially after long hours of grinding. Recently I also came across a site in which you can buy digital clothing for me it was really weird but as I saw it really becomes popular. U basically buy clothing and you can add it to your camera and it looks real. I think is good for people posting on media because thanks to it you can have a really weird and awesome drip that anyone can and you don’t need to wear some weird things. So I think we actually are starting to live in the future. Great blog 🙂