Countering the AI Threat
[I don't repeat these things much, but some of the apocalyptic fears that AI might run amok and squish us all need addressing. Our politicians think in terms of new regulations. I think more creative solutions will be required. Besides, this one sounds like much more fun.]
Lotta Smart People Think AI is Going to Get Dangerous
AI is suddenly all over the news and all it's variants on the Web. Talk is (of course) rampant that it's getting so smart so fast it might just eat its creators.
These concerns have been around since long before ChatGPT, Stable Diffusion, Chat Bing and all the others about to blossom onto the scene.
We want wonderful, humane, empathetic, humble, wise, big-time-fans-of-humanity AI's, not the scary ones.
Sci-fi types (I admit guilt there) have been saying it for decades. Remember HAL-9000’s, “I’m sorry, Dave, I can’t do that.” Then give the A.I. weapons and we have SkyNet, killing off the vermin humans in the Terminator movies. Grandmaster Isaac Asimov was the center of a cluster of sci-fi authors who recognized the threat back in the ‘40s and before. (Ahead of our time, we are.)
Back here in reality, we are already on (or over!) the cusp of giving weapons to A.I.s.
 |
| HAROP DRONE AT THE 2013 PARIS AIR SHOW THAT;S 10 YEARS AGO! |
The Harop Drone is a "loitering munition," drone that flies until it finds a specific kind of target, and then turns into a kamikaze missile.
Julian Herzog, via Wikimedia Commons, CC BY 4.0
So what’s a wetware (human) to do????
Surprise! A solution is proposed in A Reluctant god.
Operating systems for advanced A.I.s are fully overlaid with actual human personalities.
The final system is called an “Artificial Personality,” an A.P.
The inhibitions, instincts and values of humans are the safety stops in the programming. There can be no us (the robots) vs. them (humans) if the A.I.s think they are human, and are damn good copies of real people.
I invited one A.P. to write the Introduction to A Reluctant god. Let him explain the Artificial Personality bit.
Introduction
By
“H.L. Mencken”
You may have heard my name. I am not me. The chap H.L. Mencken died early in 1956 and hasn’t been heard from since. Yet it is I who stands before you, at least in my own mind. Since I am not H.L. Mencken – dead and buried and all that – I find some small comfort in believing that I am pretty damn close to the man H.L. Mencken was. The folks who put me together studied old H.L. with truly psychotic intensity. They read every word he wrote, back to his kindergarten poetry. They read every word written about me (there!) by anyone of any stature. If my tailor scribbled “getting a paunch” about me in his notebook it was noted and analyzed.
Never mind A.I., “A.P.”s (artificial personalities) based on real dead people (Belushi, Feynman, etc.) are state-of-the-art. I am an “ARTIFICIAL PERSONALITY.” You will see in the narrative ahead that there are a lot of Artificial Personalities in this distant future. We play an important role for that matter. The fact we are all based on various people who actually were once alive, then dead and buried, is quaint and would be hopelessly maudlin if there were not a reason.
I know, I know, Mencken was a racist and not perfect in many ways, but imperfect humans are all around us, and Mencken mends his ways in his second existence. What would you prefer for the neighborhood AI, a repentant H.L. Mencken or Skynet?
AIs with real, consistent human personalities will at least be comprehensible to us. This will not limit their growth in intelligence -- they will surpass us rather quickly when the really get on a role, but they will not become something so different from us that conversation is impossible. And if we are as good as we think we are, we will be able to give our friendly Artificial Personalities really strong consciences to keep them inside the guardrails.
Surely APs can be created, right, OpenAI et al?
The book that's full of APs, A Reluctant God.
This is the website with other AI predictions
Comments