Mind and Iron: An OpenAI executive just revealed how they'll replace creativity
Plus AI and the presidential debate. And a low-tech solution to all this high heat?
Hi and welcome back to another fine episode of Mind and Iron. I’m Steve Zeitchik, veteran of The Washington Post and Los Angeles Times and lead hype man of this newsy music festival.
If you're new here, every Thursday we aim to give you all the intel on where tech is taking us, no corporate spin or cults of personality. As of this issue it's been a full 12 months that we’ve been at this — Mind and Iron launched on June 22, 2023 — which means we’re beginning our second (!) year of bringing you all the news to help you navigate the hairpin future. If you haven’t done so yet, please consider a pledge to support our humanist mission. Even the lowest amount of less than $6 per month goes a long way.
This week, the scorching temperatures that have lately been burning up much of the United States both pose and reflect the risks to our survival. But a small adaptation can give big consequences — if we embrace it. Also, an OpenAI honcho says the quiet part out loud about what AI will do to creativity and jobs. We break down this revealing rhetoric to understand what the company aims to do with our expression.
And, finally, with a presidential debate afoot, where the candidates stand on AI — or even might be coached by it.
Quick housekeeping note that we're out next Thursday for the Fourth, back at ya the following week with all the delicacies your newsy palate could want.
First, the future-world quote of the week:
"The opportunities for heat mitigation in the US are huge. The knowledge is there, but the things that need to be done are not being done. It is extremely frustrating."
—Steffen Lehmann, a UNLV professor and director of the Urban Futures Lab, on what we’re doing — or not doing — to make our planet cooler
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
OpenAI wonders who needs creatives; the road from hell is paved…in white?; AI as debate tool
1. A FEW YEARS AGO, WHILE WORKING ON A STORY AT THE POST ABOUT WRITING AND GPT-3, I INTERVIEWED THE OPENAI EXECUTIVE MIRA MURATI.
Murati at the time was the senior vice president of research, product and partnerships for the (then) mostly do-gooder organization. I was struck by someone whose very careful speech concealed a more aggressive philosophy beneath. Murati and her team were clearly plotting wholesale changes in how we communicated, but wanted to be very, very careful in how she, well, communicated that.
In the years since, Murati has become chief technology officer of OpenAI (after a very brief stint as its chief executive). And she has remained just as guarded, even as she, like the (now unabashedly profit-seeking) company she works at, has been plotting ever-bigger changes to our communication and creativity.
Murati last week took the stage for a Q&A at Dartmouth, her alma mater, centered on the impact of AI. It was a truly astounding few minutes, and I think unpacking them could tell us a lot about what the giants of this field have planned for us.
As she addressed what we've been covering for a few months now (including last week) — AI allowing us to go from video curators to video creators— she first stayed on Pollyannish message. “I expect it's going to make our creativity expand,” she said in one typical comment starting at about the 28-minute mark. So far, so boilerplate.
Then she added another set of thoughts.
“We see [creativity] as sort of this very special thing that's only accessible to these very few talented people out there. These tools actually lower the barrier for anyone to think of themselves as creative and expand their creativity. So in that sense I think it's actually going to be really incredible; you can extend the story; the story never ends."
Let’s break this down, shall we?
“We see [creativity] as sort of this very special thing that's only accessible to these very few talented people out there.”
The reason for the limited “access” is because creativity takes talent, or at least work. You would never say “we see shooting a basketball as sort of this very special thing that’s only accessible to people like Luka Doncic.” There’s something — I won’t say disrespectful, but OK, disrespectful— in Murati’s attitude toward people who toil in creative realms.
In fact it goes beyond that on the brazen-ness scale, framing the whole issue as a matter of equity. “Like, isn’t it unfair that someone can do _____ creative thing and you can’t. Let Big Tech come in and even the scales. You won’t even have to try very hard!” A fine line exists between democratization and not believing in talent or sweat in the first place. Actually it’s not that fine a line. It’s bold. It’s just being erased.
“These tools lower the barrier for anyone to think of themselves as creative and expand their creativity.”
The tell here is “to think of themselves as creative.” On one level you can say Murati is just continuing the theme of “empowering people” (with work they never had to put in). But she also is tipping something else — that LLMs may not actually deliver on all this creative promise. They will just make you feel like they have, and isn’t that enough? (for OpenAI to keep moving product)
“So in that sense I think it's actually going to be really incredible; you can extend the story; the story never ends.”
I’m confused. If the product really lowers the barriers so drastically — if it gives the lie to the idea that only some people can be creative — then why would you just be extending the story? Wouldn’t you have AI write the beginning and middle too? This seems like a Freudian slip — “of course we don’t actually think it will create the next ‘Game of Thrones,’ but maybe it will give you the ending that makes you a little less angry?” (By the way this is another fallacy: what the people who don’t like a particular fictional ending primarily want is not to write their own ending but for showrunners they’ve put all their trust in to have synced up with what they want. A personalized AI ending won’t solve for this.)
But then, even the idea that it's writing a fresh ending is a bit of a grift. Because the current AI couldn’t do any of this without all the human creativity it’s based on; an LLM, by definition, never wrote anything original, it just rehashed all of someone else’s originality. It would be like saying a blender doesn’t need fruits or yogurt to create a smoothie, because see, “the liquid comes right out of the blender without any of those ingredients!” Good luck trying to drink from a blender that just mashed up air.
It continues to confound that, even as these models are hoovering up so much creative work — indeed OpenAI is running all over town making content deals by the minute — LLM companies are then disrespecting that very creative work and the creators behind it. Like, the only reason you can do what you do is because humans did it. But somehow humans don’t need to be doing it?
Murati vamped a lot about the employment automation issue. "A lot of jobs will change, some jobs will be lost, some jobs will be gained,” she said (a statement that reminded me of The Pretenders’ equally nonsensical but far more melodious “something is lost, something is found” from Hymn to Her). And she auto-played the cliche that “You can imagine a lot of jobs that are repetitive, that are just strictly repetitive [going away.]” This is by now a familiar Big Tech spin, pretending that there’s no reason for alarm because AI is just a helpful tool to replace office drudgery, and not, you know, an attempt at an entirely new form of creative intelligence.
But Murati’s doozy — the line that got her a bunch of notoriety — was this:
“Some creative jobs maybe will go away. But maybe they shouldn't have been there in the first place, if the content that comes out of it is not very high quality.”
!!
So now tech executives are a Sundance jury, telling us which creativity is and isn’t worthwhile? And somehow their main credential to be able to do this is they know how to mash it up??
It takes a remarkable feat of pretzel logic to simultaneously boast of a technology so life-changing that it can replace millions of humans and then seeks to excuse such replacement by saying those humans weren’t really doing anything special in the first place.
I won’t say that AI can’t be a tool for creators (it can, and will) or that the target of such tools won't be some lower-skill jobs. But don’t let that fool you into thinking that that's ALL the people behind this tech want it to be. Our recent industrial history makes us vulnerable to the bias that only the most rote jobs are what machine-minded execs are coming for. It’s a trope; don’t fall for it.
How much tech companies will succeed in their bid is hard to determine. Like I said, at least some of these promises of originality really are built on air. And the current stewards of creativity aren’t going to take any of this lying down, as we've seen with numerous Guild fights and saw again this week when record labels sued a couple of AI startups that create music off text prompts.
But it's important to hear executives like Murati when they quietly tell us that their ambition isn’t really a helpful new Excel spreadsheet. When it comes to creativity, AI is an overpromise (it relies on human ingenuity a lot more than its purveyors allow) but it's also an underplay, with the goal for it do a whole lot of creative human labor; that, after all, is how these companies will make money. And these few minutes of panel fodder reveal that.
So let’s translate the comments from one of the top figures at the leading AI firm into a few bullet points —
*"Creativity is really special, so why not reduce it to something as mindless and common as swiping an app?”
*AI is so powerful it will render much of human creativity moot, but don’t look at us as the culprit, we’re just providing a tool for an assembly line.”
*This technology shows how vestigial human creativity really is, which is why we worked so hard to acquire all human creativity.”
Actually Murati's comments may have inadvertently disproved her point — they may have showed how essential humans really are. After all, an AI could never have come up with thoughts this self-servingly tortured.
2. IS THE FUTURE OF CITIES WHITE ROOFS AND ROADS?
Climate change is a massive problem with a hundred potential tiny solutions that hopefully together will at least...make the problem not so large.
That’s especially true in cities, which with all the buildings and relatively few green spaces tend to get hotter, a phenomenon known as "urban heat islands." Anyone who’s spent time in one hit by a heat wave these past few weeks — NYC has felt like Hell’s Kitchen even if you’re in Brooklyn — knows how disruptive and unhealthy it can be. Also, deadly — a European heat wave two summers ago took the lives of more than 60,000 people.
One of the simplest but oddly still niche ways to defeat these heat islands is simply to repaint roads, roofs and other large exposed surfaces white — often shorthanded as a "cool pavement" program.
The notion has been bandied about for several years now. According to the studies, if you treat the asphalt with a white coating more akin to the color of the sidewalk, the reflective white (as opposed to the absorbent black) will lower the temperature on the street surface or building exterior. This not only alleviates the worst effects of heat of nearby humans but reduces the need for air conditioning and power, in turn aiding the atmosphere further. An environmental virtuous circle.
The movement began in earnest 16 years ago when a Berkeley study found that if the 100 largest cities globally replaced all roads and roofs with lighter-colored materials it would offset as many as 44 metric gigatons of greenhouse gases — a higher amount than the world emits in a year. That kicked off a wave of urban experiments, particularly in the American West.
The city of Los Angeles tried coating some neighborhoods with a lighter seal starting seven years ago. Vegas is working with casinos to coat roofs in white. Phoenix has been steadily growing its program, adding 20 percent more cool pavements last year.
The results have been mixed, but still welcome; a report last summer from Los Angeles found the lighter roads really cooled the surrounding area in the late afternoon and early evening, when asphalt streets would normally be releasing heat stored all day. (On average studies find temperatures in such cases drop by 10-12 degrees.)
One problem is while the overall effect is a cooling one, it does make people on that exact part of the pavement feel hotter, since some of the heat is reflected back on them. A study by Arizona State University researchers found that the temperature right above the white pavement was about four degrees hotter — but noted it wouldn't matter since you would be unlikely to be standing in middle of the street for long anyway and would feel the benefits everywhere else. Still this limits the applicability; you wouldn't want to use this stuff on a playground.
Maybe the most promising current research comes from an Australian effort earlier this year to figure out how to cool Riyadh. The tragedy in Saudi last weekend, when more than 1,300 pilgrims on the road to Mecca died due to heat-related factors, has underscored the importance of such measures, particularly in epically hot countries. The study from UNSW Sydney found that if all 3,323 buildings in Riyadh were optimized for cooling, including with “super cool materials implemented in the roof of the buildings,” they could actually reduce the temperature in the city by more than ten degrees.
The roof-road lightening is the kind of small change that would have outsized impacts; while we're used to blacktop roads and there is even a cultural resonance to them, changing our landscape so that we see a lot more white everywhere wouldn't really be that big of a tradeoff. And true, treating a roof costs money, but not as much money as the AC bill from pumping in all that cool air during heat waves. Faced with this choice a few years ago, executives at the MGM Grand in Las Vegas redid their roof to white and haven't looked back.
Now, driving impacts haven't been studied; cars are generally designed for (and eyes accustomed to) darker asphalt. The black stuff also better absorbs heavy rains and other corrosions. And white roads can get darker with dirt over time anyway.
But the innovation’s biggest obstacle is really just a historical resistance by some environmental experts to “adaptation” — the idea that instead of trying to mitigate the fossil-fuel emissions that cause the heat, we simply tweak our ability to live with it.
Many climate scientists, though, have started to recognize that both approaches are now needed if we hope to cool the planet, and any fears that adaptation will reduce the urgency for mitigation are drowned out by the fact that matters are desperate enough that we simply need to throw every possible solution at the problem.
Small innovations like making roads and roofs lighter are sitting in plain sight, but not being grabbed at for a host of political or petty reasons (or really just inertia.) "The opportunities for heat mitigation in the US are huge," Steffen Lehmann, a professor at UNLV and director of the Urban Futures Lab, told the BBC earlier this month. "The knowledge is there, but the things that need to be done are not being done. It is extremely frustrating." Ironically it might take a string of 95-degree days to finally shake people from their torpor.
3. FINALLY, WITH THE BIG TRUMP-BIDEN DEBATE TONIGHT — I WOULD OTHERWISE HAVE NO IDEA IT’S HAPPENING, SO IT’S A GOOD THING CNN HAS REMINDED ME WITH AT LEAST 16 SEGMENTS EVERY HOUR — I’D BE REMISS IF I DIDN’T INCLUDE SOMETHING ABOUT AI AND THE TWO MOST RECENT U.S. PRESIDENTS.
The topic will no doubt come up as one of the policy areas Donald Trump and President Biden debate tonight. They’ve both shown their hand on the topic. Biden’s executive order from the fall goes about as far as a presidential issuance can go. Which isn’t very far at all. But it’s on Biden’s radar and he’s aware of the concerns (not least because he’s been victimized by a deepfake).
I’m not sure AI regulation is going to be much on the mind of Trump, who has revealed a personal infatuation with the tech, at least as it allowed him to troll the humans it could replace (all "these wonderful speechwriters" will soon be "gone," he recently said after noting he liked an AI speech written for him).
So there will be daylight between them, with Biden likely talking more about responsible innovation and Trump unloosing the restraints on business. Which will put some of the usually Democratic-leaning Silicon Valley into a position of quietly rooting for Trump, especially given how crucial a time for AI the next four years will be. Sure, these companies have armies of lobbyists that could swat away even the loosest of regulations, but why pay them a retainer if you don’t have to?
Of course, there’s always the possibility that Trump could get badly dinged by a deepfake and turn on AI. Odder vendettas have happened.
Just for kicks, I decided to turn to AI for a general preview of what we might see tonight — what arguments we might hear advanced, what cases made, etc. The last year has brought a whole new swath of “argument generators” — programs like HyperWrite and DeepAI’s “AI Debater” that let you input any argument and it returns a rundown of the case for and against.
Generally these programs won’t give you much info or reasoning you wouldn’t have gotten from the average ninth-grader — that is to say, useful only if you know nothing about a subject and don’t have time to learn before the dinner party. Still, I put these two contentions into DeepAI’s program just to see what would happen: “Donald Trump will be a great president in his second term” and “Joe Biden will be a great president in his second term.”
It spit out this for the former:
And this for the latter:
If only the candidates could make their cases with this little rancor.
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. Last year ended with a score of -21.5 — gulp. Can 2024 do better? The first six months have been OK enough — or at least a lot better than last year.
OPENAI’S CTO WONDERS WHO NEEDS HUMAN CREATORS: -3
COOL ROADS HOLDS SOME REAL PROMISE, IF ONLY CITIES WOULD EMBRACE THEM: +1.5
EXPECT THE USUAL PRO AND ANTI STANCES ON AI REGULATION FROM THE DEMOCRAT AND REPUBLICAN AT THE PRESIDENTIAL DEBATE: 0