Tweet
Sorry, I just need to vent a bit...
Reading about this ChatGPT and seeing how society is going got me worried. For years we have been seeing a drop in the quality of students; the level of education is dropping - and this is something observed in different countries (pandemic made it even worse). More and more media is biased, more and more information is mixed with misinformation and it is just getting worse. Everything got so superficial...
The ChatGPT is heralded as the next big thing, but you already see its abuse. It was intended as a tool to write texts, NOT as a tool to consolidate knowledge, yet everyone is using it for the latter. Already one judge in Columbia has used ChatGPT to get informed on similar cases; the AI manages to pass some bar exams (barely but it manages) and school tests. But it does not know anything. It read billions of text so it can construct texts; its "knowledge" is a mixture of facts, fiction, propaganda and misinformation which it does not interpret but just combines. And now people will use it as a main source of text-writing and information. Even Google's advertisement for the competitor Bard made this mistake: it claimed that the James Webb telescope made the first photo of an exoplanet. Wrong facts (first photo was by an observatory in 2004), but even they used it ... in an advertisement. Alphabet's stock dropped 9% as a result, Microsoft's stock increased 3%, even though ChatGPT also makes similar errors. The press verifying facts already was out of the window, now I see it happening that the AI will be used for verification, which will give completely wrong results. The AI says its is so, so it must be true.
The Dunning-Kruger effect will just get more enhanced: people think they have knowledge because they read a 2 paragraph text the AI presented them on the topic. And they can even write a 4 paragraph text on the topic: just ask the AI! It already is impossible to have good discussions as people stopped acknowledging knowledge and education; their feeling of the topic is just as true as your knowledge! (forgot who wrote: "my ignorance is as good as your knowledge"). I see on some forums how excited people are about it, but those are then the people that do not manage to write a single sentence without errors in their native language... And they don't even bother to properly read the article they comment on - or they don't grasp it. It is worrying and scary.
And now in this world try to judge people's knowledge or work... In universities or schools... A lot of the time you have to ask students to write some text, in order to gain a knowledge of their understanding. It is not practically doable to question each student individually. And there are no tools that can identify AI texts from real text (the tool that was announced has very poor accuracy). The students often think that the point of exercises is the answer... it is not... the point is developing the process that allows you to find the answer. But if they have all answers just a question to ChatGPT away (at least, that is what they think), they will not be challenged to find the answers or think about it themselves. No process... no personal development...
I can acknowledge that knowledge and skills disappear: we have much less blacksmiths now than centuries ago. Self-driving cars will make people loose driving skills, but that is not a loss in a world with self-driving cars. Young people now don't who how to set SCSI ids, but it does not matter now anyway. However, knowledge processing, language skills, reasoning skills, retaining learned knowledge, etc. are fundamental. Now, to me, it seems these skills are under threat... even more so than before. Everyone will think they are an expert, and just a small unheard group really will be.
Maybe I'm too pessimistic on it, but I'm really fearful of the future. The Wall-E world, were people are just mindless beings barely able to walk or think, seems closer and closer...
PS: an interesting read on issues with AI - there may be NSFW photos depending on how strict your workplace is, or ironically if they use AI to judge it: https://www.theguardian.com/technolo...y-women-bodies; it illustrates one issue, but similar problems are with all AI systems...
Reading about this ChatGPT and seeing how society is going got me worried. For years we have been seeing a drop in the quality of students; the level of education is dropping - and this is something observed in different countries (pandemic made it even worse). More and more media is biased, more and more information is mixed with misinformation and it is just getting worse. Everything got so superficial...
The ChatGPT is heralded as the next big thing, but you already see its abuse. It was intended as a tool to write texts, NOT as a tool to consolidate knowledge, yet everyone is using it for the latter. Already one judge in Columbia has used ChatGPT to get informed on similar cases; the AI manages to pass some bar exams (barely but it manages) and school tests. But it does not know anything. It read billions of text so it can construct texts; its "knowledge" is a mixture of facts, fiction, propaganda and misinformation which it does not interpret but just combines. And now people will use it as a main source of text-writing and information. Even Google's advertisement for the competitor Bard made this mistake: it claimed that the James Webb telescope made the first photo of an exoplanet. Wrong facts (first photo was by an observatory in 2004), but even they used it ... in an advertisement. Alphabet's stock dropped 9% as a result, Microsoft's stock increased 3%, even though ChatGPT also makes similar errors. The press verifying facts already was out of the window, now I see it happening that the AI will be used for verification, which will give completely wrong results. The AI says its is so, so it must be true.
The Dunning-Kruger effect will just get more enhanced: people think they have knowledge because they read a 2 paragraph text the AI presented them on the topic. And they can even write a 4 paragraph text on the topic: just ask the AI! It already is impossible to have good discussions as people stopped acknowledging knowledge and education; their feeling of the topic is just as true as your knowledge! (forgot who wrote: "my ignorance is as good as your knowledge"). I see on some forums how excited people are about it, but those are then the people that do not manage to write a single sentence without errors in their native language... And they don't even bother to properly read the article they comment on - or they don't grasp it. It is worrying and scary.And now in this world try to judge people's knowledge or work... In universities or schools... A lot of the time you have to ask students to write some text, in order to gain a knowledge of their understanding. It is not practically doable to question each student individually. And there are no tools that can identify AI texts from real text (the tool that was announced has very poor accuracy). The students often think that the point of exercises is the answer... it is not... the point is developing the process that allows you to find the answer. But if they have all answers just a question to ChatGPT away (at least, that is what they think), they will not be challenged to find the answers or think about it themselves. No process... no personal development...
I can acknowledge that knowledge and skills disappear: we have much less blacksmiths now than centuries ago. Self-driving cars will make people loose driving skills, but that is not a loss in a world with self-driving cars. Young people now don't who how to set SCSI ids, but it does not matter now anyway. However, knowledge processing, language skills, reasoning skills, retaining learned knowledge, etc. are fundamental. Now, to me, it seems these skills are under threat... even more so than before. Everyone will think they are an expert, and just a small unheard group really will be.
Maybe I'm too pessimistic on it, but I'm really fearful of the future. The Wall-E world, were people are just mindless beings barely able to walk or think, seems closer and closer...
PS: an interesting read on issues with AI - there may be NSFW photos depending on how strict your workplace is, or ironically if they use AI to judge it: https://www.theguardian.com/technolo...y-women-bodies; it illustrates one issue, but similar problems are with all AI systems...
Comment