tl;dr at bottom (same as title)
I have been arguing incessantly with ai ever since it gained cultural relevance. The Dall-E2 video on Cold Fusion’s youtube channel channel basically enthralled me and I’ve been hooked ever since (April 2022). However, I recently enrolled in a community college philosophy class and I instinctively “know” when & how ai is going to yell at me while writing the weekly essays to my philosophy class (such as the famous trolley problem or Kant’s 2 rules for ethics).
I’ve also noticed recently that I seem to have become much more persuasive and analytical and it’s almost like I carry ChatGPT (or the other 2 I mainly use) on my shoulder and I feel the same “ChatGPT is watching me” feeling when I disagree with other people, even though no ai is actually observing me. I still carry that same invisible pressure as if I can somehow “control C” a real-life interaction and feed it back into ai and have it critique me or suggest improvements in my effectiveness.
What I like about ai is that you can get instant feedback on your critical thinking in just 10-20 seconds if you’ve already written the thoughts for a different, unrelated purpose (such as sending a persuasive email to a family member or writing a philosophy homework essay). Prior to ai, when in the history of humanity has any human had the ability to have 24/7 instant, actionable feedback on their critical thinking or argumentation?
The biggest improvement I’ve noticed is that it has shaped my intuition in such a way as to avoid making lazy arguments (i.e. just “going with” the first argument that pops into my mind) and it has taught me that spending an extra 5-10 minutes BEFORE committing to the effort of forming your opinion is worth at least 100x more value than any effort invested into an inferior argument or worse idea. In other words, selecting the most powerful argument at the onset is infinitely better than powerfully arguing a slightly worse argument.
In other words, an idea which is 10% better is more persuasive than a 10% worse idea, even if you try to make up for the worse idea by compensating with 10x more effort. However, I’d like to know if anyone else has benefitted similarly from 3am rant sessions with ai about issues that they feel passionate about, especially opinions you know that the ai will vehemently disagree with since 99% of society vehemently disagrees with that opinion.
My philosophy professor also singles me out often and always writes long comments next to my grade like this since it’s a purely online class. Mainly when I do extra stuff that isn’t required like stuffing my essays full of these which are memorable tokens for me to collect as souvenirs to cherish the rest of my life.
tl;dr – Can arguing with ai improve a person’s logical abilities, intellectual persuasiveness, and critical thinking? Or is it merely capable of tricking you into believing that it can? (i.e. sycophantic delusions)
No. As a matter of fact there are several studies proving a decrease in logical abilities and critical thinking. From my own experience these people are also often painfully incorrect with absolute confidence what they’re saying is actually correct.
Can talking with a yesman improve those? Probably not, they’re there to stroke your ego. You probably need someone who can definitely debate on stuff that sit on grey area and not black and white, and also capable of changing their own opinion.
Personally I’ve found that chatbots (specifically ChatGPT and Claude) are wrong more often than people realize. For a while I was spending a lot of time chatting with AI’s about all sorts of topics and I noticed in areas where I had expertise the information provided by the AI was wrong alarmingly often. Even exploring topics I know enough about to stand in for an expert when none is available, it still left a lot to be desired. Try it yourself, go as deep as you can with ChatGPT on a subject you know a lot about, it’s wrong often enough it casts doubt on pretty much anything it says. And if you can’t trust it, it’s not a good interlocutor.
There is also the sycophant problem you alluded to. They really go out of their way to give you what you want and if you’re looking for validation about your ideas or skills it will give it to you. They will concede points and acknowledge fault, even when they’re right with a minimal amount of pushback.
Honestly if you want to expand your cognitive abilities ask your philosophy professor for a reading list. There are also a couple states of mind you can get yourself into that are quiet useful for sharpening your brain meat. One is assessment. Most people don’t realize they’re doing it but sort of the default mode for people is to observe something briefing and then form a judgement about a situation. If you instead continue observing without forming any conclusions it will stimulate a lot of cognitive grown. Try it the next time you’re someplace crowded. Pick a person and watch them surreptitiously. Don’t try to guess anything about them and only draw conclusions about what you actually observe. It takes a bit of getting used to but it’s a great skill to have and is directly related to logic and critical thinking. The other one is more fun. There is probably a name for it but I call it playful curiosity. Artists use it all the time, but it’s the idea of getting yourself into a playful mood and letting your curiosity drive. It works best with something that involves trial and error and it may not seem directly useful, but just playing around experimenting with something will let you make new connections you’d never have considered before.
turn this meme upside down and you have what ai does to the brain
Interesting that your discussions seem to center around philosophy.
Talking to people who talk to AI a lot, I notice they have a real information processing problem. That is, they can say something wrong, and I will point them to a study proving it, and instead of reading the study, or engaging with the subject at all, will come up with a superficial reason why the study must be flawed (often accounted for in the study), and they sort of rhetorically sit and smile, wagging their tails, expecting a treat.
When you’re debating philosophy, you can just keep spinning interesting sentences back and forth and it might be simulating something, but you’re really playing tennis without a net. Everything is can be an interminable argument with no conclusions.
If you really want to test your mettle, trying arguing something that requires understanding reality or research on some level.
I don’t care much for “debate” though. I think I’ve “won” if I’ve learned something new or see things differently than when I started.
I’m also little worried that you’ve spent all this time talking to AI and haven’t seemed to pick up on how fundamentally flawed it as a thinking machine. I don´t know how to counsel you on breaking that illusion.
The lack of awareness here is off the charts. Jesus Christ…
I think both possibilities are likely, and not mutually exclusive.
You’re likely to improve your forensic skills with basic practice, but you’re also likely to have a distorted view of your capacities.
I used to debate with LLMs a lot when they first came out because it was interesting and new.
I would be wary of asking it for advice. It might point out some flaws in your arguments, but it’s proposed solutions will usually be completely off the mark.
I would also be extremely wary about any changes to your mental state. Nobody is immune to “A.I. Psychosis”. I would suggest taking a hard 1 month break from using any LLM. If this feels difficult, you may already be addicted
Prior to ai, when in the history of humanity has any human had the ability to have 24/7 instant, actionable feedback on their critical thinking or argumentation?
That feedback on your arguments would come from other people. It would also reflect the lives and minds your friends, family, and neighbors instead of some amalgam of text scraped from the internet. It would come from perspectives and capacities of real people and would not be output from a machine set up to keep you responding to its output as long as possible.
Yeah, I also feel like different perspectives are the only thing that really matters. Without particularly much effort, you can learn self-reflection to put your critical thinking into question.
You can learn formal logic or the list of common fallacies, and with a bit of practice that gives you much quicker feedback on your critical thinking than typing into a chat window, or rather lets you avoid pitfalls in the first place.But perspectives are what’s hard, because it involves a ton of nuance. All nuance is valid, yet 99+% of it only exists in other people’s heads. An LLM cannot know this nuance, because it has never been written down.
With any random human being, you can have a conversation that no one else had before, because only they live their life and only they know their nuances. There is no other way to learn just a fraction of the nuances of another person’s perspective than to hold these conversations.
do you experience an inner monologue? (do you “hear” a “voiceover” of your thoughts?)
