Dark Star and AI in Devices
John Carpenter has a thing or two to say about the use of AI in smart devices
John Carpenter’s first movie Dark Star was first released 50 years ago. John is better known for Halloween and quite a few other movies. However, I want to talk about Dark Star because this movie raised issues that are still relevant today. It’s also hilariously funny1 and if you have never seen Dark Star, stop reading and go watch it. You can find it on several streaming services for $3, which is pretty cheap these days.
Spoiler Alert: The rest of this post will make references to the movie. If you didn’t listen to me and go watch it, this is your final warning and I take no responsibility at all.
There are many amusing scenes in Dark Star. The eponymous ship is on a mission to blow up unstable planets in solar systems that might later be colonized by humans. The crew have been doing this for quite a while (20 years) and are a little space crazy. This may be due to the loss of the ship’s entire supply of toilet paper or to the acquisition of aliens along the way.
When a planet is found that needs to be blown up, smart bombs are used. One assumes this is intended to make the task more accurate, although I’m tempted to think it was because the CEO of the bomb company had bought into the recent Generative AI hype.
When Bomb #20 is armed by mistake, but refuses to go back into the bomb bay, one of the crew attempts to reason with it to prevent it from detonating. This does not go quite as well as expected. With seconds before the detonation, the bomb decides that it needs to think about its existential state and its philosophical approach. It also decides that all inputs are false and that it is the center of the universe.
If you watched the last clip or remember the movie (from when you stopped reading above to avoid the spoilers), the bomb decides to follow its prime directive and explode. Where this ties into today’s AI frenzy is clear. Do we really need AI of this sort in bombs, toothbrushes, toilets or many other devices? Or are we just following a fad?
The current approach seems to be “just because it can be done, it should be done.” To me, this may not be the most sensible approach and I’m not sure John Carpenter thinks it is either. If any of you know him, I’d be interested to find out what he does think. Yes, he is still alive, so have at it!
All movie nerds should read the article referenced above and note that Dan O’Bannon, better known for Alien among other things, worked on the script. It is not the best movie ever, but it was the start of some illustrious careers.
I think during the first generation of a new technology, "because it can be done" is a perhaps unprofitable but still useful learning exercise.
When the web suddenly became a big thing in 1996, I remember poking fun at the fact that every radio and TV commercial was suddenly advertising a website even though it really didn't matter. toothpaste.com? Yeah, Proctor & Gamble has owned that domain since 1995. I think there were even websites for toilet paper at the time, now that I think about it. But who cares? Who honestly goes to a website like toothpaste.com when everything you need to know about the product is printed on the product packaging? It seemed to be the latest, strangest fad that someone would actually connect their modem and pay AOL or their favorite ISP up to $14/hour to look up toothpaste!
And then there was the Network Time Protocol (NTP). I remember that if you really wanted to, you could sync your Windows 98 clock to NTP so that you would never need to reset for daylight savings again. But who cares what time it is down to the microsecond? Just because you could sync up clocks, you had to build an entire service for it? (I'm not done with this example, we'll come back to this in a moment...)
What we're seeing in AI is the birth of a new industry. Not all of it will be useful, but all of it will develop the skills and knowledge necessary- in the workforce to enable the next generation of this technology. Just as toothpaste.com has today been mostly abandoned, there will be casualties and underwear gnomes (summary of the joke - 1. steal underwear, 2. ??? 3. Profit! -- they never figured out step 2), but there will also be useful technologies that could not have existed if someone hadn't taken the first step.
Remember NTP that I mentioned? Without time synchronization, it wouldn't have been possible to issue time-sensitive RSA tokens in the mid-late 1990s, which has evolved into the various OTP technologies we use for MFA today, which are still dependent on NTP.
So, "should we, just because we can?" Not necessarily... but maybe.