As mythological metaphors go, the one that SNHU professor David Humphreys gave during a recent talk about the eruption of technologies known as artificial intelligence wasn’t terribly reassuring.
“We’ve opened a Pandora’s box,” he told two dozen university students during a breakout session at the College Convention hosted by New England College earlier this month. “It’s not going to go away.”
You will recall that in Greek mythology, overly curious Pandora opened the box she had been told to keep shut only to release sickness, death and a host of unspecified evils into the world. Similarly, a host of evils have been unleashed by the arrival of large language models like ChatGPT and other software products that use predictive algorithms to create astonishing simulacrums of human thought.
These evils range from the annoying, such as crummy e-books swamping Amazon, to the alarming, such as “deepfake” videos and voice recordings of people that can fool even their loved ones, to apocalyptic possibilities like AI-controlled, weapon-carrying drones.
And that doesn’t count the possibility that AI will be used to change your job for the worse in the name of increasing investor returns.
The one upside of Pandora’s story is that she also released hope. (EDIT: A reader pointed out after this ran in the Monitor that she actually didn’t release hope; she kept it in the box. Apparently this has resulted in centuries of debate over cause and effect.) The equivalent upside for AI is the way it is supercharging good things such as medical research, weather forecasting, scientific analysis and the ability to spot activities like illegal fishing and human trafficking.
Because Humphreys’ talk was part of a multi-day session about politics, he focused on the way AI can be used to fool people. “With AI it has just become easier and easier and easier to create disinformation,” he warned.
Then he showed pictures and audio of famous people doing and saying silly things, which he had created in a few minutes using online software that anybody can subscribe to for just a few bucks, as well as fake videos supposedly from the war in Gaza. Not long ago they would have required days or weeks of work by people trained on complicated software; now they can be created by bored teenagers or bad actors paid by political opponents or foreign governments.
Unfortunately, Humphreys said, there’s no easy solution, and one may not even be possible. AI-spotting software holds out some hope, but I’m very dubious that it can stay ahead of ever-improved fakery any more than spam filters have killed spam. The incentives for misbehavior are much greater than the incentives for enforcement.
As for laws, good luck with that. Unless the U.S. takes China’s route and creates total federal control of the online world, there will always be ways to sidestep even the strictest American legislation from overseas.
That leaves things, unfortunately, up to us. The old rule that you should be suspicious of things you see on the internet is now 10 times more relevant. (Note: My columns are an exception. Always believe them.)
“We need to hold each other accountable. If you’re on social media and you see people sharing misinformation, call them out. … Think before you click the share button. If you think, ‘Yeah, this is crazy!’ – it might be. Double-check … especially the more inflammatory things, especially as we get closer to the presidential election,” he said.
As you might suspect from a professor speaking to students, especially a professor whose job includes teaching other teachers how to use AI in their teaching, Humphreys is a big advocate of education. He thinks digital literacy should be part of the curriculum even in elementary schools since we’re now a society addicted to our screens.
He urged the students to “follow watchdog groups like Newsguardtech.com and Snopes” and to apply the CRAAP test to things found online, which stands for currency, relevancy, authority, accuracy and purpose.
“Look for all of those things. … The ones that are missing more than one element have a stronger likelihood of being misinformation,” he said.
I will go further and say that AI has changed the balance so much that your default for online information should shift from “it’s true unless shown otherwise” to “it’s fake unless shown otherwise.” I may be biased since this attitude gives more credence to traditional media like the Monitor, where you’ve got real, live folks to hold accountable, but I’m afraid it’s a necessary step. AI is just going to get more powerful, and it’s not like we can turn it off.
“Is it worth it to shut the box?” Humphreys asked in a rhetorical moment. “Can we deal with the bad things, to get all the good things that come along with it?”
The answer, of course, is that we can’t shut the box. But we can minimize the damage: “If we can teach people how to effectively and ethically share information online … then we’re doing what we can do to help prevent the spread of mis- and dis-information.”