on ai praise
Inspired by Ava's shortsighted AI praise II.
Note: By "AI" below, I specifically mean generative artificial intelligence (generative AI) using large language models (LLMs) to generate text.
My understanding is that AI isn't conscious and doesn't share wisdom like humans do. All it does at a rudimentary level is predict the next token (a word or a piece of a word) to output based on the input.
Better than us
What does "better" mean to you? Who is "us"? Anyhow, AI doesn't need to be better than us to be better than someone on topics like conversations and advice.
I want to make some points on "the best of humanity". AI can help someone without picking from that. Training data can contain helpful insights without including that. Humans often don't pick from that either in their own communications.
Even though I praise AI, there are many times where using AI as a tool is worse than direct human involvement. The opposite is also true.
When I think of "AI praise", I think of praise for not only humans' technological marvel, but wielding AI to solve problems they otherwise would have solved poorly or not at all. I'm sure there are more breakthroughs to praise too.
AI benefits me
I think people with lacking or no social support would be bettered by repairing that. Even though I think this would be best done through humans, I think crediting the AI when it helps somehow is warranted.
I'm familiar with the due diligence and associated benefits of compiling insights from strewn-out sources of information. Whether from material sources (forums, books, or social media) or human sources (therapy, strangers, supportive friends and family, etc.), this can take a while and may not leave me satisfied.
I find AI exciting for how it synthesizes those sources of information into one avenue. Yes, I can acknowledge AI generates sentences derived from human communications, although some training data sets even includes artificially generated data. That doesn't stop me from seeing how AI can save people time, money, energy, and countless Internet searches.
Sometimes, AI has given me better responses for certain asks than I found elsewhere. I'm grateful to have access to those, as well as other information I wouldn't have had access to. In a sense, I see AI as an augment to my own imagination. Yes, this can even includes its hallucinations at times.
On data privacy
No, I don't believe AI should be mandated. Like other contentious features commonly embedded in programs and services, it should ideally be opt-in.
To be clear, I'm aware that AI services and tools including AI can be prime targets for privacy and security incidents. You should think twice about what you publish, prompt for, and even use with how your information can be disseminated nowadays.
For those wanting data privacy with their AI usage, plenty of trustworthy local AI software solutions and smaller LLMs can be run on consumer graphics cards. There are even numerous LLMs small enough to use on smartphones and other portable devices. You can even interface AI without an Internet connection if you prefer.
None of this is to say that I should prompt or otherwise contribute to non-local AI with abandon. There are pieces of information I would be best served keeping to myself. What I'm getting at is noticing more nuanced and acceptable use cases for non-local AI. I can do that while acknowledging that it and its local counterpart each have their own strengths to exploit.
On idea ownership
However, concerning my personal AI usage, thinking through AI discourse has me reconsidering the value of non-local AI.
The more I contribute toward training data, the more my words may help people prompting AI for relevant output. I can make parallels to how learning in public blasts lessons outwards, instead of hoarding them.
As an example, I don't have any preventative measures against AI scrapers. If they haven't scraped my blog yet, I'm sure they will in due time. If my communications help one person better their life through that, I see how that's worthwhile even without credit.
Instead of worrying over maintaining inventor status, I can encourage superspreaders to proliferate my work. Concerning my blog again, I have two main desires from readers like you right now. First, email exchanges to co-create conversations and other works of genius. Second, remixing my work as much as possible, in the same spirit that I'm inspired to write from many other bloggers around me. The more I post, the more credit and attention I may naturally receive based on how prolific I am.
Every stolen work of mine that I find is an opportunity. I can capitalize on it through gifting appreciation, inviting collaboration, or even learning how I'm underselling my own product. Adapt press for your purpose.
Unexpected similarities
Don't humans "copy humans while being created, trained, and curated by humans"? Mimicry is embedded within conversation and other human interactions. Communication, knowledge, and social norms get influenced by the "training" of life experiences. Humans not only curate other humans, but curate responses to others based on various factors too.
Pragmatism
Giving technological tools like AI, smartphones, and other commonly condemned beasts your responsibility is a dangerous game. Learning how those are tools you can use to sculpt reality can unshackle you from your self-imposed handcuffs. Even the people maximally enriching their lives through technology, instead of casting them off as bipolar totems of fun and distraction, have room to grow.
Yet, I can only influence people so much to redirect their responsibility back to themselves. My energy may be better spent on those who already decouple themselves from their tools this way. Is there anyone out there?
Want to reach out? Connect with me however you prefer:
- Email me via your mail client
- Copy my email address or remember it for later:
yoursimperfect@proton.me
- Email me via Letterbird contact form or open it in a new tab