Dear Friend,
It’s been a little more than four years since I wrote Disrupt Disruption – How to Decode the Future, Disrupt Your Industry, and Transform Your Business. It’s been a super fun ride, and the book laid the groundwork for OUTLEARN and REWIRE (the next one in the Built for Turbulence series). To celebrate the occasion (and because information wants to be free), we made the eBook of Disrupt Disruption available for free – no strings attached. Download it here in PDF and ePub/Kindle format – and feel free to share it with your friends and colleagues.
P.S. Here is a lovely BBC clip from 1964, featuring Arthur C. Clarke talking about his predictions for the future. Shared by friend of radical Manos Xenos.
And now, this…
Headlines from the Future
The AI Value Gap: Enterprise leaders love to gloat about their AI investments – and say much less about the value they are getting from their investments. The AI Value Gap dashboard shines a light on the issue – only three in ten large companies can point to a single realised, quantified AI benefit. The website provides indexes, rankings, and an AI Claim Finder. This is one of those benchmarks/dashboards you want to keep an eye on.
The Waymo Effect: Daniel Hook, the Chief Scientific Officer at Holtzbrinck Group (the German publishing company), argues that AI is making research less collaborative. Just like riding a Waymo removes the sometimes awkward small talk with your Uber driver, an LLM removes the awkward colleague who tells you you’re solving the wrong problem – in both cases we feel only the relief, and miss the potential benefit of being challenged in our thinking. By automating tasks like literature reviews and writing, researchers may bypass the deep thinking required to produce high-quality work. Meanwhile, the current academic incentive system rewards this rapid, solitary output over the slower, more costly process of human partnership, and an LLM never argues about author order. This is a challenge which is not limited to academia but something affecting every organization which is measuring output as a key metric.
AI vs Human Context Windows: A paper by Netanel Eliav from the Machine Human Intelligence Lab explores “Cognitive Divergence” – the growing gap between AI’s rapidly expanding ability to ingest and process information and humans’ decreasing ability to do the same. In simple terms: AI’s short-term memory is exploding, while humans increasingly struggle with focus. Eliav argues that these aren’t two independent trends: the more we delegate, the less we practice, and the less we practice, the more we delegate. In sixteen years, the time a person holds attention on a screen before switching away dropped from about 150 seconds in 2004 to about 47 seconds by 2020. This widening gap creates asymmetric access to knowledge: AI can analyze massive amounts of information, but humans no longer have the focus or patience to check the source material behind the AI’s answer. All of which calls for future AI and educational tools to be designed to scaffold human thinking rather than completely replacing it.
Kevin Kelly on Predicting the Present: Here’s a lovely collection of quotes from the first five years of Wired compiled by Silicon Valley legend Kevin Kelly on the future (i.e. our present). Many turned out to be utterly wrong (MIT Media Lab’s Nicholas Negroponte in 1993: “I expect that within the next five years more than one in ten people will wear head-mounted computer displays while traveling in buses, trains, and planes.”), many are spot on (our friend Paul Saffo: “The scarce resource will not be stuff, but point of view.”). Worth a read – they tell you a lot about what 1990s Silicon Valley believed about itself.
AI as a Psychic’s Con: Baldur Bjarnason argues that LLMs operate on the playbook of a psychic – through the use of statistical genericism and validation statements, they trick the user into believing they are capable of genuine reasoning. Bjarnason concludes that relying on these models for serious decision-making is equivalent to seeking advice from a psychic hotline. Personally, I am not sure I fully agree with this – even if LLMs don’t possess genuine intelligence, the complex mechanical machinery they are built on allows them to operate in genuinely smart ways. And yes, don’t rely on LLMs for complex decision-making.
AI Sucker Punch: Online betting site DraftKings deployed AI to help it identify and target customers most likely to lose money – and egg them on. A New York Times investigation found that the company is using advanced machine learning algorithms to prioritize promotional incentives for “elastic” bettors (the ones who lose the most for every free bet they’re handed) – while sidelining projects aimed at detecting addiction risks. Both models analyze the same betting records and identify the same people – but as a former employee said, “the best investment would be a problem gambler.” DraftKings chose to pick the one that makes it money and buried the one that doesn’t.
Surviving Survivor: The TV show “Survivor” has been around since 2000 – the premise is simple: Get a bunch of people on a remote island, make them do a bunch of stuff, have them vote each other out, and then let the ones they voted out pick the winner. In the end, it’s a social strategy game. Data scientist Victoria Ritvo developed an AI model to predict who would win the show – and it picks the winner about twice as often as chance. Aside from the fun of using machine learning to predict human behavior in a game show, it gives us another glimpse of what’s possible when data meets human behavior (see the DraftKings example above to get a sense of what happens when you use massive amounts of data to do something similar).
What’s It Like inside a Data Center: Data centers are a hot-button topic these days, but what does it actually look like working inside one? The Wall Street Journal digs into the topic and finds (not all that surprisingly) a lot of pretty banal stuff, carried out by a small group of employees.
Agents That Run Your Company: We are still some ways off from having agents run (even smaller parts of) our companies, but that doesn’t keep people from trying and experimenting. Andon Labs had agents run anything from vending machines to a coffee shop – and they want to do it for you too. We recommend treading carefully here – but it’s fascinating to see what the tip of the spear in AI looks like.
Where Are the AI Killer Apps: In his New York Times op-ed, Paul Ford points out that although AI has made it so much easier to code apps, we don’t seem to have a killer app (yet). I think that’s a fair point – but it also misses a bigger point: I’ve been running small AI-coded apps of my own, and a few have become killer apps for a market of exactly one: me. Maybe AI’s killer app is not the app for the masses, but the one app for you.
What We Are Reading
Kara Swisher: AI Isn’t Going to Destroy Humanity - but the People Building It Might Forget the doom and the hype – Kara Swisher has a much simpler explanation for what’s going wrong with AI, and it starts and ends with the people in charge. @Jane
Meet the Guys You Call When AI Breaks Your Brain AI is driving some users into full-blown psychosis, convincing them they’re geniuses, fostering delusional relationships, and leaving lives destroyed. A 25-year-old house painter from Quebec is running a support group to help them cope. @Mafe
We Surveyed 634 Women Who Work in Tech. They Let Loose Part of WIRED’s special Women issue, this survey captures a workforce embracing AI while still navigating many of the same cultural barriers the industry has promised for years to leave behind. @Kacee
Dating Apps Are Dying When abundance becomes a problem: Dating apps are dying – not because people don’t use them, but because too much choice in dating removes all the friction which makes falling in love possible in the first place. Maybe we are looking at the canary in the coal mine here. @Pascal
…and now for some fun.
🚘 Los Angeles’s Highway 101 is notorious for its traffic. Turn the never-ending gridlock into a relaxing soundscape where each car, recorded from live traffic cameras, plays its own tune.
What We Do When We’re Not Writing
Hi! I’m Pascal from radical. When we’re not writing this newsletter, we help organizations turn volatility into advantage – without the “innovation theater.” We build your team’s capacity to handle disruption, so you stop reacting and start shaping. If that sounds useful, let’s talk.
↗ Dive into the deep end: Access our complete collection of 3,000+ radical links.