Part of the fun of Star Trek, for me, is how it teases the futurist in me.
When I was a kid, I tried to figure out how the Transporter would work in real life. I thought about warp drive, and whether or not it could be real. That pulled me into physics, and got me reading all kinds of books.
And, of course, much is made of how we now have real-world communicators, phasers, and tricorders. And computers we can talk to.
And, of course, there’s the AI.
Early Trek had some interesting AI: androids that could pass for human, computers that could run entire planets. But these were cheesy, even clumsy attempts at presenting future tech, based on as-yet poorly understood concepts; AI was, after all, an idea that had only really been around for a decade or so when the original Trek premiered. But Trek’s AI got better and better as the years passed and new incarnations appeared.
Now it’s 2023 and some of our favorite Trek characters of all time, like Cmdr. Data, are artificial. Captain Jean-Luc Picard himself now has a synthetic body. And the Picard Trek is serving up some fascinating variations on the theme that are tickling the futurist in me once again.
AI Minds and Android Bodies
The pacesetter for artificial minds in Trek is Data, of course, from Next Generation. A sentient android able to work with and form relationships with his crewmates, Data is so human-like that he aspires to full humanity. Data’s ‘positronic’ brain is downloadable, creating the option of moving his mind between android bodies.
This occurs upon Data’s death in Star Trek: Nemesis, when he transfers his artificial mind into the body of an earlier prototype of himself, B-4 – a move that is less than perfect, as B-4 is a lesser model.
Android bodies? We got ‘em. They aren’t yet ubiquitous, but they’ve been around for a while and they’re getting better every year. For instance, there’s a prototype named Atlas, built by Boston Dynamics in partnership with the Defense Advanced Research Projects Agency, which can drive industrial vehicles, walk across a terrain covered with rubble, run across a clear terrain, open doors, climb ladders, use tools, find and manipulate valves, connect hoses, and so on.
Not only are the digital brains of existing robots amenable to update, they are Internet-friendly; they can have their brains stored in the cloud. Their design allows for the digital brain of one to be loaded into the physical body of another.
Put another way, we’ve hit this mark already.
Human Minds as AIs
What about downloading human brains into computers or android bodies? Per Trek, this is doable: Dr. Ira Graves, mentor of Data’s creator Noonian Soong, downloads his own mind into Data’s body just before his death. When Jean-Luc Picard dies, his mind is downloaded into a ‘golem’, a synthetic simulacrum of a humanoid, by Soong’s son Altan and Agnes Jerati.
In the latter case, the transfer of Picard to a synthetic body involves “copying” his mind before his death, which we witness; the Picard “copy” is what makes it into the golem, in the end. The copy is not the original, and that’s a pretty big deal, but we won’t deal with it here; we’ll note that the end result is a synthetic humanoid with a mind of human origin that thinks exactly like the original, behaves exactly like the original, and even believes itself to be the original, even though it’s just a copy.
But the more interesting possibility is this: Robin Hanson proposes the emulated mind (EM), an AI created by scanning the brain of an adult human in excruciating detail and creating a perfect digital model, replicating it precisely, that could then be run as a computer simulation. This would be, as above, a digital mind of human origin with would think and behave like the original. Hanson proposes that EMs would capture human expertise and competence brilliantly and give human-level competence to the computer-automated, robot-powered world that is already rapidly emerging around us.
This is basically a variation on the Picard scenario, though there is no illusion that the scan that becomes the model is in any way conscious; it is simply a repository with perfect fidelity. On the other hand, Hanson points out, EMs, once created, could be endlessly replicated. This would utterly redefine the economy of the Earth, to say the least.
Hanson believes this is inevitable, and will occur during this century. The progress of both computer technology and brain imaging techniques is already barreling in that direction. Once we can do it, we surely will.
Not here yet, but getting close.
Digital Minds in VR
Between his dying body and the golem awaiting his mind, Picard is temporarily accommodated in a virtual reality based on the den from his French chateau. He experiences the virtual environment as if he were a physical human being in an actual room, from the flames in the fireplace to the comfy sofa.
Data is there with him – or, rather, a copy of Data, who experiences the VR simulation just as Picard does. The two digital copies of the original Picard and Data share one last moment together before Data requests that Picard erase him completely. Once Picard’s digital incarnation is transferred to the golem, he grants Data’s wish.
I don't need to go further than my daughter’s Oculus headset rig to enter a virtual world that is breathtaking in detail (though I was lucky enough to experience US military VR simulators in the early Nineties, making me one of the early birds). VR is already here, and our military already uses it to train human beings and pit them against AI-based digital adversaries within the simulations.
All that’s left to achieve the Picard scene above is to replace the human player in that current-day VR scenario with one of Hanson’s EMs, and we’re there.
Again, we’re getting close – but only to a simulation. We could put them in the room, and they could have their conversation, but neither would be self-aware.
Human and Digital AI Minds in the Same Body
In a final scenario, a later Picard scenario includes a Soong golem, such as Picard has become, inhabited by 1) a copy of Data, 2) a copy of Lore, 3) a copy of B-4, 4) a copy of Data’s ‘daughter’, Lal, and 5) a copy of Alton Soong, presumably created before his death in the same manner that he copied Picard’s mind.
Can that work?
It’s already happened once, when the copied mind of Ira Graves cohabited with Data in Data’s positronic brain. But five digital minds at once?
This takes us into an architectural terrain. If we view the brain of the Data golem as a VR environment, then five occupants is a technologically viable event.
But this raises the specter – also an issue in the Chateau Picard VR scenario above – of the nature of the hardware. We won’t deep-dive it here, but the fact is that there’s a vast difference between a mind (human or AI) running as a simulation in a digital computer, and an actual machine consciousness, which necessarily runs not as a set of programs, but as a connectionist system – an actual physical neural network, rather than a digital simulation of a neural network. We’re told Data is designed with such a network, and we can assume the same for the golems; but it’s unreasonable to believe there could be five digital minds operating at the same time, running on one physical neural network not designed for anything else.
In other words, this might be possible, but it won’t happen anytime soon.
Ah, well; I have more than enough to keep me happy during my remaining years, I think.
Commentaires