I Have Seen The Future, And It Dislikes Us.
That’s a lot worse for tech than it is for us - maybe.
Full disclosure: I work with several tech companies doing interesting and valuable things with AI, and all are focused on customer needs. They are exceptional, and survivors. I’m careful who I work with.
The Artificial Intelligence apocalypse may not arrive as advertised. Unless AI companies can figure out how to enjoy thinking about their business customers, and learn to make things people want to pay for, the failure to find and please customers is going to kill a lot of them. It could be bad for everyone else, too.
I spent much of last week at a big tech gathering in San Francisco, with thousands of software developers and startups showing off their latest AI marvels and hammering out their dreams of a new future, full of new gadgets and ways of making new tech companies and billionaires.
While much of it may well be amazing, world-changing stuff, I came away thinking that most of it will never see daylight, due to one the oldest problems in the tech industry: For the most part, tech companies don’t really like mainstream business.
At a session on selling AI to corporations, panelists talked about all the ways customers are a problem: They have security concerns; No one can figure out how to use their stuff; There isn’t enough internal maintenance of the new AI tools; People are suspicious that AI will take their jobs; Executives weren’t using the right metrics to measure success.
When the seeming problem is the customer, dear bros, the real problem is you. That’s an age old lesson, one I’ve seen tech companies fail to learn for three decades.
When I started at Google Cloud, I heard stories about sales people (really, genius engineers tasked with wooing customers) who would basically say, Our stuff is the best, so you’re stupid if you don’t use it. For all the hand wringing over how Google has lost its soul, Cloud has filled its ‘plexes with professional sales people who can at least feign empathy for ordinary businesses, and has since raised its market share.
Twenty years ago Nike spent $400 million on supply chain software, then lost $100 million when its people couldn’t work with it the way the tech supplier required. People who work in tech at mainstream companies swap stories about the problems caused by hostile tech products. During the Internet boom, someone at a futuristic company described his outfit to me as “the nimble mammals, dancing around the dying dinosaurs.” His low-empathy company later went down from a lack of business.
Despite those lessons, last week an AI founder told a room that top executives at prospective companies “get it - they want to be more productive while employing fewer people.” Ten minutes later he complained that employees don’t want to use his product because they think it will take their jobs. Go figure! Elsewhere, a founder said his product was so amazing that he didn’t need to sell, people would simply come to him. Another offered an AI avatar managers could use to appear at meetings they didn’t want to attend personally. Literally, enabling them to take skin out of the game.
Of course, my survey was unscientific, but these attitudes rang a familiar bell, and I now think that AI’s alienation from its potential customers is a big problem. It could hit the real economy almost as hard as it could hit the tech sector, too, since much of the current U.S. GDP growth (and hence the stock market and interest rate policy, among other economic factors) is tied to capital investments in Artificial Intelligence. AI is going to have to offer returns on that investment in a timely way. Economist Noah Smith published an excellent piece on this problem last weekend.
On the face of it, this timeless issue in tech is utterly crazy. For all of the industry’s talk of disruption and transformation, tech counts on the existence of a mainstream economy doing mostly mainstream stuff. E-commerce is now about 16% of all retail, but it took decades to get there: The first e-commerce company, Boston Computer Exchange, started in 1982. All but a vanishing morsel of the trillions thrown at new digital technologies over the past few decades has been invested with the underlying assumption that new tech will land in a still-larger economy, where consumers have disposable incomes.
If this is to happen, tens of thousands of non-tech companies must become customers. Personal tech like the iPhone is flashy, and changes things up relatively quickly, but the reality is that mainstream businesses (which for some reason tech companies refer to with the antiseptic term, “the enterprise”) are much more important to the bottom line. Consumers spend something like $1.4 trillion on tech, globally, but companies spend $4.5 trillion. In addition, it costs a lot less to sell to companies, compared to the expensive ad campaigns of consumer tech. Supposedly boring normal companies are actually the lifeblood of Silicon Valley.
Microsoft may not have the greatest reputation among techies, but you have to say this for it: Bill Gates built a culture that cares about boring. He liked to fly to Omaha and eat Big Macs and talk business with Warren Buffett. The stereotypical Microsoft person is a square who likes golfing with tire manufacturers and talking tires, or catching a beer with cocoa salesmen and discussing commodities. And deep down under that folksy exterior is a very hard, competitive human, respectful of places where it’s tough to succeed.
Unlike Microsoft, most tech companies don’t like normal business. Maybe it’s because people in tech think they’re smarter, or that what they are doing is much more interesting than boring old business. Maybe it’s because the net profit margins of mainstream companies (about 8.5% overall - groceries are 1-2%) are less than half that of an enterprise software company (19.5%), a gap which perhaps makes tech companies feel like they’re in a different and better league.
Or maybe it’s because techies think they are inventing the future, and the status quo is, by definition, heading into the past. The part smug techies miss is that the future generally arrives consisting mostly of the present, and changes a couple of tweaks at a time. When the future arrives suddenly, it usually means a meteor has hit the Earth.
Which, in tech, is often a selling point: This changes everything! And this time the scenarios from AI CEOs are unusually apocalyptic: AI will send unemployment to 20%; it will cause the extinction of humanity. But then again, given the hundreds of billions in funding for AI, maybe they need commensurately apocalyptic forecasts.
“No jobs and everybody dead” would, in most industries, seem like a negative. In enterprise tech marketing, though, promises of big disruption can excite fear (your competitors are jumping on this lethal thing!) and greed (you can win markets, fire staff, and drown competitors, making your stock explode!) Can’t get more basic than that.
Disruption also makes it possible for tech companies to sell hard-to-use products, which often force workers to act more like computers than people. Don’t worry if this new expense software is frustrating garbage - when you learn to use it you’re disrupting yourself!1 Much easier than figuring out who customers are and making something that makes sense to them.
When CEOs talk about wiping out all the jobs, while they don’t seem to even begin to understand what those jobs and businesses are about, tech arrogance is back to its worst ways. Having the world throw billions at your science project can do that. Even so, your science project only exists within capitalism, which at its best is about pleasing customers.
Not to compound the big bummer here, but last week I also saw something else that had me wondering about the prospects for AI in the economy, which I’ll write about next.
There is a substantial industry that’s mostly built around getting people to use new technologies. It’s called “change management,” and more often than not it’s based on the human reality that people hate giving up the stuff they know how to do. Google Cloud, for example, had a very hard time getting people to give up Microsoft Office, a communications system few enjoy, for Google Workspace, which includes Gmail, Google Docs, the calendar, and other stuff. Pretty much everyone uses these in their private life (the spreadsheet was a change, but not major), so they already knew and used these products. Even so, they were paralyzed by the idea of adopting them in the workplace, and looking less than super competent for the couple of days they’d need to get used to them. I’m not kidding, learning a new tech system is a real deal killer.
This very human reluctance is not what I’m talking about here. I’m talking about rather obtuse tech tools that fail badly at relating to normal people. I once devoted a column to the impossibility of even reviewing a new product by the company that made once-ubiquitous Blackberry communicators, because the company put me through so many hoops, synchs, updates, and requests for personal data before I could get the thing to even turn on.
I wrote, “I hate myself for not being good enough for my technology.” I bet you know the feeling.
Challenging ideas here. But I sleep better on my Luddite pillow.