In draft, titled “Should self-driving cars individually learn?” but on thinking about it, the real issue is what sort of Borg will self-driving cars be?
This Wired article raises great questions (now that driverless vehicles, or at least semi-driverless vehicles, are among us). For example: should autonomous vehicles each learn individually (humans learn as individuals) or should they have their “AI brain” updated globally from the Borg collective hive mind (i.e., from self-driving software vendor HQ)?
The AI systems available today are not intelligent. Not even remotely – I mean compared to human intelligence. But suppose some were. Suppose some could intrinsically learn an yield human-like results. Should the AI learning systems be installed in self-driving cars? Or should the cars be mere destinations for global online updates from the Borg Central intelligent computer system that does the actually learning?
Or should there be a combination: an initial load from Borg Central, then learning by the vehicle itself yielding different knowledge, to some extent, per vehicle?
A third option is that vehicles transmit what they individually learn to other vehicles. Alternatively, individually learning could be transmitted to Borg Central, the hive mind, then Vendor HQ could transmits this collected “learning” back out to many individual vehicles (all getting the same “learning”).
From memory, the Borg hive receives data from individual Borg, then updates that information (somewhat processed) to all other individual Borg. Borg do not (if I remember right) swap data between themselves.
So what will self-driving cars do once AI systems can actually learn (in the human sense) about their surroundings?
Well, the shareholders of self-driving vehicle software suppliers will demand monetization. The cars won’t be allowed to learn for themselves. Their data will be transmitted to Borg Central and will be the input for learning by the hive mind. Then the centralized learning will be transmitted out to all the cars of the same model whose owners have paid the maintenance fee.
Pretty amazing prospect. Pretty scary too! What if that process gets hacked?
And that’s not the worst of it.
Rumor has it that at the advent of cellular phones, the spies were ecstatic. They realized that a few dozen of their Christmases had come at once – phone conversations and meta data would now be far easier to secretly intercept and record than previously.
The really scary prospect for AI that can learn in a human-like way, is that Borg Central is just a nick name. The real names are NSA and GCHQ. Borg Central might actually be Spy Central. The spy agencies might tap into transmitted self-driving car data the same way the NSA reportedly tapped into Google user’s data.
Then Spy Central could have a real-time view of everything going on around every self-driving vehicle. That really is what the Borg were all about.