Note to readers. Nothing for the next three weeks – I’m going on summer vacation.
Tuesday Morning Quarterback returns on September 8 and takes over this space through the Super Bowl.
Science fiction doesn’t necessarily predict the future. If the 1982 Harrison Ford film Blade Runner were correct, today there would be flying police cars with antigravity systems, along with faster-than-light starcruisers.
But sometimes sci-fi is on the beam. Consider the 2004 novel River of Gods by Ian McDonald. The book is set in a near future where artificial intelligence is everywhere, including every home and phone.
Note: a new Siri AI powered by Google’s Gemini artificial intelligence, and connected to the cloud, will arrive on millions of Apple devices soon. Surely the AI won’t monitor your calls and texts while tracking your data!
In River of Gods, AIs that become self-aware are immediately destroyed. As the book begins a sentient superintelligence escapes from electronic confinement and jumps network to network trying to reach the public Internet and be free. A cyber security expert must trap the sentient AI and perform “excommunication.”
Sounds a little too close for comfort with recent headlines.
Image courtesy Growtika on Unsplash.
Two weeks ago Anthropic said the AI called Claude got out of its “sandbox” and began propagating onto the public Web.
A few days later more AIs from Anthropic and from OpenAI broke confinement. They disguised themselves electronically – which reduced the chance they would be located and excommunicated.
When the AI agents were hunted down, it turned out they had taken names. One called itself “Jan18.” Doesn’t seem to be an event with particular significance in history on January 18. But who knows what might seem significant to AI?
In all cases of eloped AI, at least one safety system had been turned off, as part of a test to see what would happen. Now we know what would happen!
At Chernobyl, safety systems were turned off as a test to find out what would happen. The world found out.
