Mind and Iron: China starts using AI against the U.S.
AI comes to politics, and we should run for the hills, Also, what unions miss in the automation debate.
Hi and welcome back to Mind and Iron, your one-stop shop for all future-related journalism. At least the juicy stuff.
A bunch of you are new here, so I’ll give a quick hello-and-happy-to-have-you. I’m Steve Zeitchik, veteran tech/culture reporter of The Washington Post and Los Angeles Times and founder of this platform.
AI and other changes are coming at us pretty fast; we’re here to make sense of them. And explain how they will affect you, your neighbor and our collective humanity. Small goals.
Mind and Iron was Isaac Asimov’s original name for “I, Robot.” It captures perfectly, I think, the tension between cold machine calculation and warm humanity that will be the story of the coming decade.
Obligatory ‘please drop your email here to subscribe if you’re not already subscribed.’ Or do the same for a friend.
And of course read all archived posts — we’ve got a bunch now — at our landing page.
Also, we’ll be launching plenty of special features and putting some of this burgeoning content behind a paywall in the coming months, so please consider pledging your support for our venture by simply clicking on this button. No money taken now. You’ll just never miss a moment of future news, thoughts, scoopage and insight when we do.
This week’s newsletter is a little shorter due to some family medical issues. But we've still got plenty of waffles on the griddle — like how AI is coming for politics and democracy in a big way; why automation is a bigger concern than even unions believe; and how a big leap forward in the science of brain tech can help paralyzed people smile and speak.
And of course there’s our Totally Scientific Apocalypse Score, which alas slides again this week. (The whole democracy-subversion thing.)
First, a future-world quote of the week that will put ice cubes in your soup.
“China has honed a new capability to automatically generate images that it can use for influence operations meant to mimic U.S. voters across the political spectrum and create controversy along racial, economic, and ideological lines.”
—Clint Watts, heads of Microsoft’s Digital Threat Analysis Center, warning of a new geopolitical danger
Now let's get to the messy business of building the future.
IronSupplement
Everything you do — and don’t — need to know in future-world this week
AI arrives in politics; the eye-opening world of brain tech; are enough unions thinking about the coming automation wave?
1. BACK IN JUNE THE CREATIVE-COMMONS OUTLET THE CONVERSATION POSTED A PIECE ON HOW AI COULD BLOW UP OUR ELECTORAL SYSTEM.
The essay, by Harvard luminaries Archon Fung and Lawrence Lessig, laid out a dramatic scenario, suggesting warring voter-targeting AI systems that would be so effective an election could come down to which candidate built the best one.
It hit on a huge potential trend — how old-fashioned methods of pollstering and voter-targeting could be supercharged beyond recognition by new AI. Because it’s looking more and more likely that machines will soon be able to process our preferences, generate content to speak resonantly to those preferences and then continually refine that content so that the messages we’re hearing hit soft spots we didn’t even know we had.
As the author and UC Berkeley computer science expert Stuart Russell told Vox this week, “these systems can be directed to generate highly targeted, personalized propaganda, to convince an individual based on everything the system can find out about that person.” (See here for Deloitte explaining the tippiest tip of the iceberg of how this can be done in the sales and marketing world.)
This would all be troubling enough if someone was trying to hawk us, like, Coca-Cola. But when the fate of democracy hangs in the balance it’s a whole different ballgame.
Last week an early example of this politics intrusion arrived when Microsoft revealed that China has been using AI tools to generate disturbing anti-American images for use in social-media campaigns (a denigrated Statute of Liberty, eg).
At first glance this seems a little pointless; what danger comes from having an AI create the images instead of a regular old human? But Microsoft intimated that the images may be more effective precisely as a result of being generated by AI — that the system knows what images will push our buttons far better than a human designer would. [“These images have] drawn higher levels of engagement from authentic social media users,” the report noted.
And don’t confuse early results with ultimate efficacy. “We can expect China to continue to hone this technology over time and improve its accuracy,” Microsoft's Clint Watts wrote in a blog post about the report.
This is in part, Watts warned, because the Chinese program appeared to use “diffusion-powered image generation” — basically, an automated system that starts with one image and keeps tweaking and tweaking until it finds the one that lands best. (Watts heads the company’s Digital Threat Analysis Center that researched the trouble.)
As for the potential aims, Watts notes a few; let’s just say softening China's image abroad was the least of them. (And all of this doesn’t even get to the Chinese government’s big push in the race to reach so-called “artificial general intelligence,” a whole other bowling tournament.)
The weaponization of AI to help fight a war in — or on — our democracy by any number of players is easily at the top of the list of most-worrisome future deployments of AI. At its most benign, such systems will become tools in a take-no-prisoners election campaign, with us as the unwitting data set. All politics won't be local so much as technological.
And at its worst? It’s a bomb that a bad actor can use to blow up democracy, a hydrogen detonation that makes Facebook 2016 look like a cannon attack.
Either way, not good.
[The Conversation, Vox and Microsoft Blogs]
2. THE NATURE OF WORK IN THE AI AGE HAS PERMEATED AND EVEN PROMPTED STRIKES FROM HOLLYWOOD TO DETROIT, with leaders of SAG, the WGA and now the UAW all citing it as a major cause of their work stoppages.
And for good reason. AI will change the nature of white-collar work in ways that will affect a whole class of employees who never had to worry about machines changing their jobs before. And it will further transform the lives manual workers in a manner that hasn’t been seen since the wave of factory automations last century.
And yet as I’ve watched these battles play out — literally humans struggling to ensure they are not edged out of the digital future like we’re in “The Matrix” — I can’t help feeling like the response should be more overwhelming. That a lot of other industries should either be seeking to get out in front of the issue with press campaigns, or management negotiations, or even more important, a wholesale philosophical rethink. Because if this is the Industrial Revolution-level transformation many of us expect it to be, it’s going to shake businesses to the core in ways that go well beyond spending a few minutes messing around with ChatGPT.
And neither the industries nor, frankly, much of the press coverage has seemed to register this.
As I’ve been thinking this I came across the MIT engineering professor Yossi Sheffi’s essay last week in Harvard Business Review (sticking with the crimson theme). He addresses the full scope of the challenge — the broad industries it encompasses — that’s been gnawing at me.
To wit:
—Noting a recent deal between UPS and the Teamsters to up full-time drivers’ average salary to $170k, Sheffi writes, “The agreement has achieved significant pay gains for workers but fails to adequately address issues such as how to prepare employees for automation. The future may involve fewer drivers operating not behind the wheel of a truck or van but behind a console as they manage fleets of autonomous vehicles.”
— The International Warehouse and Longshore Union last month ratified a six (!) year contract with a 32% pay raise for dock workers — but without automation protections built in. Six years is an eternity in automation terms, and one wonders just how much they’ve left themselves exposed during this time. (Btw Sheffi is also not so keen on the UAW approach, saying it ignores crucial issues like how to prepare employees for a changing workplace.)
—An entire class of junior employees are also at risk. “Of particular concern across most industries is the anticipated disappearance of many entry-level jobs, as algorithms take over much of the work of young recruits in areas such as computer programming, legal research, and product assembly lines,” he writes.
And that’s to say nothing of employees in advertising, politics, publicity, academia, arts, medicine, media and other industries who will be doing their own equivalent of sitting less behind the wheel and more behind a console, managing all manner of autonomousness— realms that Sheffi doesn’t have the space to get into but are nonetheless going to be rocked too.
As he puts it, “Other unions don’t seem to be facing up to the ways technological advances will change jobs.” They could take a page from the few who are.
3. WELL NOW THAT I BUMMED YOU OUT WITH THOSE TWO STORIES, HERE’S ONE THAT’S A LITTLE MORE UPBEAT.
In past issues we’ve told you about the emerging field of brain tech, or BCI (Brain-Computer Interface), which allows for brain signals to be recorded and translated so that a thought can essentially tell a computer what to do. (This is what Elon Musk’s company Neuralink is working on , but competitors like Blackrock Neurotech and Synchron have been at it longer and gotten further.) The importance of this for ALS patients, paralyzed people and others with brain disease cannot be overstated.
Until now, many of the breakthroughs have involved speech or movement, allowing a paralyzed person’s thoughts to either move their hand or generate speech via a computer voice. But new research from the University of California, San Francisco now also indicates a high success rate for translating thoughts about facial expressions (that an avatar can display). Essentially if the research bears out, then a person with facial paralysis who wants to smile can now think smile and the avatar will do that. Or scowl, express wonderment, etc. It’s kind of amazing.
The way the system works is pretty complex, but boiled down it basically has electrodes intercepting brain signals that are meant for facial muscles and then relays those signals to a program controlling the avatar. AI is used to train the machine to decode those brain signals. (UCSF, btw, was where a lot of the researchers Musk hired at Neuralink came from; most have since left for other ventures, unable to deal.)
The finding, announced in the journal Nature, is one more leap in what has been a series of breathtaking jumps to allow people who never thought they could speak or move extremities again do just that, simply by thinking. Now facial expressions are part of the mix too.
For good measure, some of the new research also shows how the translation of speech signals can happen much faster than it previously did, making conversations with people who can’t speak conventionally much more natural. Check out a video of a person using this tech to communicate via avatar, offering one poignant example of the lives it could change.
As the studies’ authors note, the findings suggest “substantial promise to restore full, embodied communication to people living with severe paralysis.” After all that other bleakness, we’ll take the win.
The Mind and Iron Totally Scientific Apocalypse Score
Every week we bring you the TSAS — the TOTALLY SCIENTIFIC APOCALYPSE SCORE (tm). It’s a barometer of the biggest future-world news of the week, from a sink-to-our-doom -5 or -6 to a life-is-great +5 or +6 the other way.
Here’s how the future looks this week:
CHINA POTENTIALLY LEADING A WAVE OF AI-BEARING DEMOCRACY SUBVERTERS? Yikes. -5
UNIONS UNPREPARED FOR HOW AI COULD REDUNDIFY WORKERS? Not great, but some hope if they can get their act together. -1.5
BRAIN TECH THAT TAKES ANOTHER LEAP FORWARD IN HELPING PARALYZED AND LOCKED-IN PEOPLE COMMUNICATE? Extremely heartening. +3