I am still fascinated with AI, partly because of the undeniable coolness of its applications and partly the way a bird is fascinated by the hypnotic movements of the snake about to devour it. It’s compelling! Will it kill us all? Or save the world? There are experts in the field yelling opposing opinions about that.
Probably the truth lies, as it usually does, somewhere in the middle. Maybe it will make some things better even as it adds to the devouring of the natural world, the degrading of habitats, the ravaging of natural systems and the dumbing down of humanity.
It’s so easy to shrug and say, “Why think when you can ask a machine absolutely anything? So what if it lies? In this age of misinformation who can tell what’s true anyway? All I want is an answer. Any answer.”
I’ve found a balance in my use of AI. I never consult it unless I can’t find the answer I’m looking for somewhere else. And I never discuss my feelings or personal life with it (unless I have a specific health related question and it’s the middle of the night).
I have experimented a little with the different models, and I have settled on Claude, from Anthropic, as by far the most reliable: here’s why.
I performed a series of small experiments using ChatGPT, Gemini and Claude. I’ve already rejected Grok for reasons that might be obvious (it’s Elon Musk’s personal AI pet).
I asked a question about something I was trying to do with my phone. “How can I jibjab the whatzername, and adjust the razzle frazzle so it works?” Something like that.
I asked all three models the same question. First I asked ChatGPT, then Gemini… For the first two, the processes were similar. They each gave me answers that didn’t work. When I said so, both replied with variants on, “Oh yes, you are absolutely right, Apple has upgraded the whatzername process so now the razzlefrazzle function has been hidden.”
I asked what the current solution was. They kept giving ever more complicated wrong responses, apologizing profusely, drawing me into a rabbit hole of frustration.
And then I asked Claude. It answered much more slowly than the other two had. It spun for a while, ‘thinking… cogitating…ruminating…’ and then gave me a simple, accurate answer. It made no further suggestions and no attempt to keep my attention.
Then I tried something different. I love webcomics. I found one quite a long time ago, all I remembered was that the main character was a female ogre, a veteran of a terrible war, who had found sanctuary with peaceful folk, and that the art was painterly in style. I couldn’t remember the name. I gave the same prompt to all three models. The first two could not find it. They kept asking for more detailed information, returning ever more unlikely suggestions, and finally I gave up.
Then it was Claude’s turn. I gave it the same prompt. It took a good 20 seconds cogitating. At last it answered, “Got it! The name is Fairmeadow and here is the website URL.” Impressively, it was correct. But most impressive was, unlike every other chatbot I have engaged with, it never tried to keep me engaged. It answered what I asked, and then it shut up.
I don’t use AI a lot. But when I can’t find an answer, I do ask Claude, and it’s nearly always helpful and succinct. The most annoying thing about ChatGPT is, it will never simply end a conversation. It always ends with a question or a suggestion for further engagement. Can it be that alone of all the AI models, Anthropic has taught its model to be helpful, end of story? Not to grab and hold onto my attention no matter what?
I haven’t decided that AI is a good thing. Data centres are undeniably horrible. The potential dark outcomes of AI on our society, on jobs, on humanity, the darkest forecasts, terrify me. But is it more horrible than what we are already doing? Fracking, shaving old growth forests bald, bad water in poor communities, wars, drones, the military industrial complex, sex slavery, rampant predatory capitalism, the pollution of our planet and our minds, the extinction of species, the list could go on and on. And maybe AI will help in the end. Somehow.
To righteously eschew AI while participating, passively or otherwise, in the other horrors of this topheavy, unsustainable society feels pointless to me.
I would abstain, but how do you unplug from a system you are embedded in?
I minimize my online activities. I hardly do social media anymore.
I go outside, I bike, I write, I sing, I talk to people, I do yoga, I think deep thoughts. I do not talk to AI about philosophy or ideas or feelings, any more, that is. I am developing my inner resources. But just as I drive a car (as little as I can), and own a phone (to which I am addicted), I am caught on the treadmill of technology.
We need to talk about grounded, community based solutions, because the technofascists are not going to do anything to fix problems they are actively creating.
In the meantime, I will fiddle as Rome burns. Fiddling is fun. I don’t feel good about it. But you have to laugh.
That’s what I think. What do you think? Email me at phoenixonhornby@gmail.com
