Dan Siroker knows how to pivot. He once had an education startup that morphed into a successful online analytics company called Optimizely. More recently he created an AI company called Scribe.ai, dedicated to capturing data from apps like Zoom, then reintroduced it as a MacOS app called Rewind designed to be “a search engine for your life.” That wasn’t clicking, so this week he changed the name of the company again to Limitless and introduced a spiffy wearable. It’s a $99 clip or pendant that can record all your conversations so you can use generative AI to refresh your memories or analyze your interactions with other people.
Siroker also has some good ideas about privacy: One feature prevents the recording of someone’s voice unless they verbally grant permission. But he also understands that while his long-term vision encompasses a deep, lifelong record, Limitless needs to solve a problem today to win its initial customers. Siroker’s initial target is the tedium of unproductive meetings. “It’s a real problem that technology can make meaningfully better,” says Siroker.
Siroker sees AI as a way to transform the endless procession of meetings, in-person or remote, scheduled or ad hoc, that professionals suffer through daily. He’s got lots of competition. Productivity companies like Microsoft, meeting apps like Zoom, and transcription startups like Otter are also layering AI on top of our meetings. We should watch what happens closely, because the way AI transforms this slice of the working life is an indicator of the future of work in an AI-powered world.
The AI meeting singularity has started benignly, with the seemingly innocuous step of transcription and summarizing. It’s the automation of a task once routinely assigned to secretaries. Even that change subtly alters the dynamic of conversations, because now all meetings have a paper trail, stripping deniability from every stupid remark that slips from someone’s mouth. Or the AI-generated transcript can lay bare which team members haven’t uttered a word in three months of planning sessions. But it’s undeniably useful when AI can quickly and accurately summarize a discussion and identify what actions need to be taken. Those features can pour digital grout into cracks that assignments previously slipped through.
“Humans forget 90 percent of what happens after just one week,” says Siroker. “If you had an hour-long meeting a week ago, you will at best remember six minutes.” Limitless uses AI to prevent you from forgetting the important parts, sending messages drawn from your data via its app to proactively prep you for your next meeting. For example, it might warn you that you previously promised to do something and report back.
These are only the first baby steps toward what seems destined to be a bigger revamp of meetings. It would be a mistake to think that the algorithms in the room will remain mere observers. AI is more like an ambitious virtual worker seeking a promotion, or at least a more active role in the proceedings. One day an AI-powered service might actually run the meeting for you. And why not? It will know better than anyone else in the room how the company works, what its immediate goals are, the status of its customers, and the capabilities of other attendees.
Not everyone sees things this way. Steven Rogelberg, a professor at UNC Charlotte, has been studying meetings for years; his most recent book is Glad We Met: The Art and Science of 1:1 Meetings. While he views AI as a helpful enhancer, Rogelberg believes that meetings as we know them will always be a vital part of work. “Humans have been gathering for thousands of years,” he says. “AI is not going to eliminate that need for humans to come together.” When he advises companies looking for AI to improve their meetings he cautions them that real improvements still rely on the human fundamentals. “That means the leader creates agendas that are highly strategic and engaging, having the right attendees at the right time,” he says. “The tool doesn’t replace that.”
Tell that to the people building systems with generative AI. I set up a call with Sam Liang, the CEO of Otter.ai, a company that started as a simple transcription service and now is focused on using AI to remake meetings. (Naturally, Otter kept track of our conversation, summarizing as we spoke.) One of Otter’s meeting tools has the ability to compare the meeting agenda to a real-time summary that ticks off which items are addressed. It would make sense, I mentioned, if instead of having the leader check those summaries, the AI itself kept things moving, butting in when it realized that people spent too much time talking about one item when there were only a few minutes left for three remaining items. Liang tells me this feature is actually in the works. I wondered whether it was only a matter of time before the AI actually moderated the meeting. Liang doesn’t miss a beat. “It could be a moderator,” he says.
It could also, he says, be a personal meeting coach, helping people perform better in gatherings or one-on-ones. For instance, if a junior person in a sales call got a tough question, the AI might provide advice gleaned from the calls of experienced closers. Or it might simply tell the salesperson to stop talking so fast. Coaching from an experienced hand is always helpful. But the idea of everyone in a meeting connected to their own personal coach feels vaguely dystopian, as if the people in the room voluntarily made themselves into puppets whose strings are manipulated by a large language model.
I ask Liang whether the prominence of AI in meetings might make humans less likely to attend. Knowing that there will be a summary available seems a disincentive to actually showing up. Liang himself says that he attends only a fraction of the meetings he’s invited to. “As CEO of a startup, I get tons of invitations to go to meetings—oftentimes I’m double booked or triple booked,” he says. “With Otter, I can look at my invitations and rank them. I classify them based on the content, the urgency, importance, and whether my presence add any value or not.” Since he’s the CEO, he may find it easier to opt out. On the other hand, the boss’s presence in a meeting makes it more valuable to those who want clues to his thinking or an instant yes on a proposal.
Of course, the premise behind meetings is that every person’s presence adds potential value. It defeats the purpose if at the moment everyone turns to the single person who can weigh in on a problem, they find only an empty seat. But Liang has an AI solution for that too. “We’re building a system called Otter Avatar that will train a personal model for each employee for meetings where the employee doesn’t want to go or is sick or on vacation. We will train the avatar using your historical data, or your past meetings, or your Slack messages. If you have a question to ask that employee, the avatar can answer the question on their behalf.”
I point out that this might lead to an AI arms race. “I’m going to send my avatar to every meeting, and so will everyone else,” I explain. Meetings will be just a bunch of AI avatars talking to each other—afterward, people will check out the summary to see what the AIs said to each other.
“That can happen,” says Liang. “Of course, there are always situations where you want a personal relationship directly.”
“In that case,” I reply. “I can go out to a bar with those people.”
“Yes, you can have a drink with your coworker while your avatars are having a meeting with each other!” says Liang. “Ultimately you don’t need a job, because the avatar did all the job!”
We were riffing now, but there’s a serious undercurrent to this speculation. We are entering a period in AI development where businesses are embedding the technology in powerful products to be used in collaboration with humans, with the flesh-and-blood contingent firmly in charge. But many of the people building the technology are fixated on a mission to build so-called artificial general intelligence that can outperform or replace humans. If all goes to plan, what begin as useful tools could take on increasingly prominent roles in the workplace, replacing at first the pre-AI way of working—and later human workers too.
At that point we can meet up in those bars, spending our universal basic income checks on drinks. Maybe we will be wearing Dan Siroker’s pendants to capture our conversations so we can add them to our ever-expanding life archives. One question that’s sure to come up: “Can you help me remember what it was like when we used to have those old-time meetings at what used to be our jobs?”
Time Travel
Boring, nonproductive meetings are one of the constant indignities of the corporate life satirized in the long-running Dilbert comic strip. In this 1996 Newsweek story, I wrote about the character exposed to maddening foibles of office culture, as drawn by a pre-politicized Scott Adams.
The central tenet of this dyspeptic corporate vision is the Dilbert Principle. As Adams puts it, “The most ineffective workers are systematically moved to the place where they can do the least damage: management.” Of course, this creates maximum damage, as their idiocy permeates corporate fife. “It seems as if we’ve turned nature’s rules upside down,” Adams writes. “We systematically identify and promote people who have the least skills.”
Every month or so in the “Dilbert” workplace, some bizarre management fad dribbles down to the drones: Reengineering, Total Quality Management or paintball tournaments. They are timewasters at best, tortures at worst. Hours are spent in meetings about deadlines, deadlines that get harder to make because of all the hours spent in meetings. Technology has run amok; the engineers understand how it works, but the bosses—who can’t tell the difference between a PowerBook and an Etch-A-Sketch—don’t get it at all. Every so often, an order comes from above to devote massive amounts of time to make everything “IS09000 compliant”; no one knows what IS09000 is. Instead of getting products out the door, people are asked to memorize mission statements. And in the background, burning ever closer, are the fires of Competition, triggering the dread drums of Downsizing. “Knock knock,” says the boss. “Who’s there?” asks the employee. The boss grins: “Not you anymore!”
If this isn’t hell, it’s close. Even Dilbert’s creator admits that those two tufts of hair sticking out of the boss’s mostly bald pate are modeled on the Devil’s horns. “Over time,” says Adams, “his personality gets more defective, and his horns get higher, making him look even more demonic.”
Ask Me One Thing
Quentin asks, “Got any advice on how to weed out real from fake advertising?”
Thanks, Quentin. This is a big problem, because some of the largest purveyors of advertising seem to have abandoned the principle that protecting customers from scams makes their own business more valuable. Did you ever see an ad on Facebook selling discounted or discontinued Forever stamps? Here’s a news flash: the United States Postal Service does not treat its stamps like outdated sports equipment, sending old commemoratives to random websites. Someone like Quentin, who is aware that advertising can be fake, probably knows this. The question is, why doesn’t Meta? I’ve seen those ads multiple times and can’t figure out why the company isn’t more aggressive in policing them. Likewise, Amazon has been struggling against a tsunami of counterfeit goods, even counterfeit authors! (A Meta spokesperson said that fakes appear on all platforms and “We invest heavily in our trained enforcement and review teams and have specialized detection tools to identify compromised accounts and other fraudulent activity.” The company’s help pages offer some tips on avoiding shopping scams.)
Some basic common sense rules: Unless you have a relationship with a seller, anything that comes out of the blue packaged as a bargain is best ignored. Anything that seems too good to be true is generally false, especially on the internet. Avoid clicking through to the websites featured on some of those ads—some of them use a misspelled or twisted version of a brand you think is what’s selling. Go to the genuine website of a brand and see if you can find the offer there. And realize that when you choose a low-priced item from an unfamiliar brand on a site like Amazon, the chances are high that you will be disappointed with the quality. At the least, take care to read the comments and reviews of an item: quite often, unhappy customers share their dissatisfaction. (Although reviews, too, can be fake.) Often you can Google an item and see whether anyone has blown the whistle on a fake. But you can go pretty far by trusting your instincts—and curbing your hunger to get one over on the seller. Or as W.C. Fields once said, “You can’t cheat an honest man.”
You can submit questions to mail@wired.com. Write ASK LEVY in the subject line.
End Times Chronicle
A “global bleaching event” is rocking the world’s coral reefs. At least it will cure them of Covid.
Last but Not Least
Will AI really put us out of our jobs? The CEO of the Federal Reserve Bank of San Francisco says it’s up to us. OK … no. Now tell that to GPT-10.
Revealed at last! The scientists who experimented on themselves to help win World War II.
Google workers are protesting the company’s contracts with Israel. Who will tell Gemini?
Is your life too pleasant? Hook up with an AI-powered angry girlfriend and be miserable!
https://www.wired.com/feed/rss
Steven Levy
