Tobold's Blog
Monday, May 20, 2024
AI stories as a lense on society and true value

It has become impossible to open a news source without coming across some story about Artificial Intelligence. AI is making progress or being proposed as a solution for all sorts of human activity. There are people who are in favor of AI, and often would like to make money with it, and others who think that AI will bring some sort of doom on humanity. My personal approach is very different: I believe that all these stories about AI shine a light on our society, and might reveal what about it is phony, and what has true value.

While AI is great at imitating humans, it isn't capable of actual critical thinking or creativity. That for me poses a rather fundamental question: If what you are doing can be replaced by AI, then how creative or intelligent was that activity in the first place? A typical example is content creation on the internet, where the prevailing thought for years was things like search engine optimization, or optimizing your content for some other algorithm, e.g. on YouTube. Surprise, surprise, an AI is in itself an algorithm, and understands how other algorithms work better than most humans do. If I would set up my blog to post an optimized article written by AI daily, my viewer numbers would go up, even if the interest to actual readers would probably go down. I am blogging to have a space to express my personal thoughts, so I am not interested in maximizing viewer numbers. But a content creator who is just after the money might very well be already a slave to some algorithm, and might be better off if he switches to AI.

I believe that interaction with an AI can at best be a mediocre experience, inferior to true human interaction. But what about human interactions that have been turned so toxic, that a mediocre and meaningless interaction with an AI is actually preferable? Dating app Bumble recently suggested that AI could date "hundreds" for you. The sad reality of dating apps is that they enable a small minority of attractive people to get endless numbers of dates, while more plain people are just fleeced for their money without getting the love the platform promised them. I don't know if there really is "a lid for every pot", but both the plain lid and the plain pot are more likely to "swipe right" on somebody more attractive, and "swipe left" on each other in this system. If AI girlfriends are becoming the next big thing, it tells us something about society, how human to human dating has become too stressful and humiliating to many people, so that an AI girlfriend now appears to be a better option.

The same thing is true for general friendly interaction with other humans: We have turned social media into toxic places filled with rage. If chatbots are the nicer alternative, that tells us more about how worthless a "friend" on social media is than about how meaningful an interaction with an AI friend can actually be. We might have reached, or will soon reach, the point where a chatbot is less likely to encourage you to commit suicide than an actual human. Oh, great! What progress!

In the end the dead internet theory might be less of a conspiracy theory, and more of a glimmer of hope for the future. Maybe humanity would be better off leaving the internet to bots and AI, while concentrating on real life human interaction with true value.

I think for starters we need to differentiate between current forms of AI and the potentials of AGI. ChatGPT can be useful if you already have an idea and can critically verify the results as opposed to just taking it for the bare truth. But it is far from an AGI which can bring doom to humanity in the same way we manipulate our environments to fit our needs.

The other day I had to write a bit of code and didn't know the correct API syntax, so I was tempted to just use ChatGPT for the research and then go from there (I found a better source, so I didn't). I think that would be fine. But I also met someone who was learning a foreign language and used it to explain the usage of words. He swore that it never made mistakes... Needless to say that his usage of the language was rather.. experimental.

When you say that AI interaction will be inferior to true human interaction, I would agree. But I would also say that it's a True Scotsman and even your average human interaction is inferior to true human interaction. It's not that they are toxic but rather bland, mediocre and meaningless (at least for me, so I rather don't).

In terms of dating apps, it's not that dating is more stressful or humiliating but rather that the tools aggregate a larger pool in which it is harder for the average person to land on top of the list. It's the old optimisation problem in context of choice overload. So people will skip local optima BECAUSE they can see further and gravitate to a better optimum.
The solution isn't to switch to AI but to drastically reduce the choices and force people to compromise (potentially under the guise of "this is your perfect match, make it work").

I think social media is that way because a) "content" creators (including vast swaths of the traditional media) are slaves to the ad algorithm and b) watching paint dry doesn't sell.
So they present stuff that "engages" and that is easily done through polarising content.
Then you can lean back and watch two sides argue over irrelevant things AND you don't have to invest into quality or reflected analysis.
It's also a downward spiral: it takes time and resources on both sides to produce as well as to consume and reflect on it. And the competition can easily jump in with some dumb clickbait and draw all the attention. So your effort was wasted.
And your competition is pretty much everyone with access to the Internet.
I’ve been enjoying your posts which I find thoughtful and informative, and thought I’d leave this comment here to see if commenting on Blogger works for me now (didn’t before for some reason).

Thank you for writing.
Post a Comment

<< Home
Newer›  ‹Older

  Powered by Blogger   Free Page Rank Tool