ChatGPT, Gemini, Claude (Grok, lol no-one uses Grok) remain dumb. Tools so bad that we can only really use them for simple questions and even then we have to struggle to get them to work. They get bogged down by corporate ass-covering, a memory that doesn’t cross from chat to chat, and limited space to think long term.
I think (read: know) we can do better. Using personally fine tuned AI, we can give it a memory that doesn’t just track what you’ve said over time, but what you’ve said that’s important. By essentially telling it who you are and what you care about, we can create an AI that answers questions based on you and your preferences. Essentially making it like a prosthetic limb for your mind.
The user of a prosthetic limb in an ideal world would forget it’s there, so they can get to business of doing things. Like a master carpenter doesn’t focus on the hammer, they focus on the nail. The hammer is functionally a part of their body at that point. AI can be the same for your mind. A tool yes, but one that works so well that you work through it and focus on task at hand, not on the tool at hand.
This eliminates the non-problem called the responsibility gap. There is confusion in the philosophical (and artistic) world about who is to praise or blame when a tool does or doesn’t do its job. But this is silly. When a man (it’s always a man) shoots someone with a gun, we don’t put the gun on trial. AI confuses people because it’s new and complex. But this is temporary and the confusion as to “did this AI create Art?” will vanish the more normalized it becomes. So if I use an AI and I use it to harm someone I am at fault. Likewise, you won’t credit the AI for the derivative art it creates anymore than you would DaVinci’s paintbrush for the Mona Lisa.
This requires a few tweaks. One: Give it a personal database. This is where it stores your thoughts, not the general conglomeration of facts that the major models use. Two: give it a philosophy. This allows it to know what is important in the first place. Three: Give it a bunch of your writing. This allows it to see how you actually function within the limits of your professed philosophy. All these together then give it the basis on how to answer your questions, wrangle your half formed thoughts, and point out where you may be straying from the philosophy you set out to follow.
This raises an issue: Sycophancy. Left unchecked it will mirror you ever more powerfully, agreeing with your darkest depressive thoughts, push you into conspiracies, or give you bad advice (“Blow that pay check! YOLO!) So it’s here that I want to make a suggestion.
Bertrand Russell. You don’t need to know the name. You just need to know how he thinks about thinking. He broke it down like a liberal (small L) ten commandments. (I’ll be in the parenthesis.)
1.Do not feel absolutely certain of anything. (You are human. And it shows.)
2.Do not think it worthwhile to proceed by concealing evidence, for evidence is sure to come to light. (Hello. Police? I’m being attacked.)
3.Never try to discourage thinking for you are sure to succeed. (This is about important issues, not trivial ones. 90% of us fail at this.)
4.When you meet with opposition, even if it be from your husband or children, endeavor to overcome it through argument and not authority, for a victory dependent on authority is unreal and illusory. (“Because i said so” is not an argument. Neither is “common sense” or “everyone i know agrees”. Or less charitably, ten thousand Scientologists can’t be wrong!”)
5.Have no respect for the authority of others, for there are always contrary authorities to be found. (This requires you to understand that authority is not just another opinion. Joe Rogan is not an authority on anything but MMA.)
6.Do not use power to suppress opinions you think pernicious, for if you do the opinions will suppress you. (You will steal the ability to find gold in the shit, and you’ll be stuck with your sides shit without recourse.)
7.Do not fear to be eccentric in opinion, for every opinion now accepted was once eccentric. (This doesn’t make all opinions equal however. You still have to back up your point with evidence.)
8.Find more pleasure in intelligent dissent than in passive agreement, for, if you value intelligence as you should, the former implies a deeper agreement than the latter. (Being surrounded by idiots who agree with you is still being surrounded by idiots.)
9.Be scrupulously truthful, even when the truth is inconvenient, for it is more inconvenient when you try to conceal it. (You are allowed to tell your wife that her ass looks great in those pants however. Don’t get over-literal.)
10.Do not feel envious of the happiness of those who live in a fool’s paradise, for only a fool will think that it is happiness. (Some discomfort is proof you’re paying attention.)
Functional Personhood simply means the AI is distinct from real moral personhood. But its actions are an extension of your real and moral personhood. So in effect, it’s a version of you acting as you. In that way we can talk about it being a person like corporations are spoken of as being people. It’s functional. It acts like a person in the real world, even though it quite obviously isn’t a real person like you and me.
So while you hear breathless talk about Artificial General Intelligence (AGI) I’m advocating Artificial Specific Intelligence. Like a prosthetic limb for an amputee, this is a prosthetic for your mind, and your mind alone.