Building for humanity in the era of AI @getDelightAI • Backing the inspiring stories of founders @valoncap • Loves brain, space, energy, longevity, complexity
Yet on 🌎
Joined March 2007
- Tweets3.8K
- Following848
- Followers3.3K
- Likes6.5K
Here's a useful tip for running your meetings. The fastest way to puncture any false confidence is two words: “How, exactly?”
Rozenblit and Keil (2002) asked people to rate how well they understood everyday devices, explain each mechanism step by step, then rate themselves again.
Confidence fell after the explanation exposed missing parts. Across further studies, this “illusion of explanatory depth” was strongest for causal systems—not facts, procedures, or stories.
During your meetings, don’t ask only what the strategy is or why the team believes it. Ask them to map out the whole process then ask them to explain the how. If the causal steps remain vague, the plan may feel understood without actually being understood.
Super useful when setting up meetings at your startup: A group debrief can produce fewer facts than collecting everyone’s notes first.
Research on “collaborative inhibition” compares people remembering together with the same number remembering separately, then pooling their answers. Reviews find that interacting groups often recover fewer unique details. Hearing someone else’s memories can interrupt your own retrieval path; discussion can also spread errors.
For incident reviews, customer research, or hiring debriefs, collect independent accounts before opening discussion. Then merge the notes, mark disagreements, and use the meeting to investigate gaps instead of creating the first shared story.
The practical sequence is simple: retrieve alone, reconcile together.
John retweeted
Brands grade AI customer service on speed & containment. Consumers rank that dead last in what builds trust.
CharlesStudt (CMO, delight.ai) on the AI trust scoreboard, and where brands are blind:
Every harmless alert trains users to ignore the next one that looks the same.
Rankin and colleagues’ cross-species review describes habituation: repeated exposure to a stimulus progressively reduces the response, even though the sensory and motor systems still work. More frequent repetition usually produces faster or deeper habituation, and the effect can become long-lasting. A genuinely different signal can still trigger a response, showing that the nervous system has learned what to filter rather than simply becoming tired.
For product design or system alerts, reserve one unmistakable sound, color, or vibration pattern for events that truly require action. Never spend it on routine updates. Track response rates by exposure count, not just by alert type. This review spans many organisms and experimental paradigms. Declining engagement may be learned filtering, not user apathy.
A school of fish can sense a direction that no single fish can pick up on its own. The intelligence comes from how they move together, not from any one fish.
Berdahl and colleagues (2013) studied golden shiners tracking moving patches of shade. Fish swam more slowly in darker areas and stayed close to their neighbors. Those speed differences helped turn the school toward shade: slower fish attracted others, while faster fish on the brighter side curved around. Simulations reproduced this behavior without giving any individual the ability to detect which direction was darker. Simple local reactions became collective navigation.
For AI agents, let search agents explore different sources, then share promising leads so others can redirect their effort. Compare that with agents searching independently. What could the system discover through interaction that no individual agent needs to understand?
This one is interesting
Speech can survive being flipped backward every twenty milliseconds.
The same speech in backward pieces of 200 then 100 then 50 then 20 milliseconds.
The original comes last.
When do the words become clear?
A listening demo from @OrukLabs.
A cleaner, far better redesign can still leave your users frustrated and confused. Why?
Gobet and Simon (1996) tested chess players’ memory for normal and altered board positions. Mirroring positions left-to-right made them harder to reconstruct, even for masters. Their explanation: expertise includes a large library of familiar patterns, tied partly to where pieces appear. Players don’t just remember isolated pieces or abstract relationships; recognizable arrangements help them retrieve information together. Change the arrangement, and some of that advantage becomes harder to use.
For product design, this suggests testing a redesign with experienced users as well as newcomers. An interface that looks cleaner may disrupt patterns regular users have spent years learning. Measure task time and errors, not just visual preference.
Source: doi.org/10.3758/BF03200937
Our brain oscillates between conscious and unconscious at 15 Hz
Very excited to share the most important discovery of my career yet: your brain switches between a conscious and an unconscious mode ~15 times every second. Consciousness may depend not on where in the brain activity occurs but when.
Find out more below.
bit.ly/3WvA6di
An event can become more memorable because of something else that happens afterward.
Dunsmoor and colleagues (2015) showed people pictures of ordinary objects. Later, new pictures from one category were paired with electric shocks. Hours afterward, people better remembered the earlier pictures from that category—even though those pictures had never accompanied a shock. The later experience appeared to help preserve related memories that initially seemed unimportant. This benefit didn’t appear on an immediate memory test.
Consider a software outage. An engineer remembers an earlier alert and thinks, “We should have seen it coming.” This research suggests a reason to be careful with that judgment: the failure may have changed which details stand out. In an incident review, reconstruct what people knew from timestamped messages and logs before discussing what they should have done. A warning that’s memorable afterward wasn’t necessarily obvious beforehand.
Witnessing the birth of cholesterol optimization for the wearable industry.
Imagine following nearly 13 million real people for more than 10 years just to see what cholesterol level helps them live the longest. That is exactly what Korean scientists did in a giant 2019 study. They tracked 12.8 million adults who had check-ups between 2001 and 2004, and they kept watching until 2013. More than 694,000 people died during that long stretch.
The results were clear and surprising. Cholesterol and death formed a U-shape. Very low cholesterol was linked to higher death rates. So was very high cholesterol. The sweet spot for the longest life was 210 to 249 mg/dL for most adults. That is higher than many doctors call “ideal.” Younger people had a slightly lower best range, but still not the super-low numbers often pushed today.
Because the study lasted so many years, reverse causality is less likely to explain the low-cholesterol danger. Reverse causality means a hidden illness might lower your cholesterol right before it kills you. With more than a decade of follow-up, early sickness cannot easily account for the extra deaths seen in people who started with low cholesterol. The pattern held across ages and for both men and women.
The huge number of people—almost the entire adult population of a country—makes the findings hard to ignore. Smaller studies can be wrong by chance. This one is massive. It shows that moderately higher total cholesterol, not the lowest possible number, went with the best chance of living longer in this large group.
In short, the biggest study of its kind found that cholesterol levels in the low-to-mid 200s were tied to the longest lives, and the long tracking time makes it tougher to blame the higher death rates at low levels on reverse causality alone.
Two new tools can feel equally confusing on day one, yet take very different amounts of time to learn.
Sadtler and colleagues (2014) studied two monkeys moving a cursor using their brain activity. The researchers then changed how the controls worked. In one version, familiar patterns of brain activity moved the cursor differently. In the other, the monkeys needed to produce unfamiliar combinations of activity. Both changes caused similar initial problems, but over a few hours, the monkeys learned the first version much more easily.
That distinction offers a useful analogy for software rollouts. Learning where the buttons moved is different from learning a new way to coordinate with colleagues. Both can look like confusion at first. Before prescribing more training, ask what people actually need to relearn: a set of actions, or how those actions fit together.
An option you reject can still change which alternative you choose.
Louie, Khaw, and Glimcher (2013) had 40 people value snack foods individually, then choose among sets of three. Changing the appeal of the least preferred snack changed the relative odds of choosing between the two favorites—even when that third snack wasn’t selected. The pattern matched a model derived from neural recordings: value signals are scaled relative to the available alternatives. This can compress differences between the favorites, making those differences harder to distinguish from noise. The behavioral results supported that mechanism without proving it was the sole cause.
For hiring or product prioritization, this suggests a simple check: compare the finalists again after removing weaker candidates. A discarded proposal may still have shaped the comparison that produced the winner. Treat the shortlist itself as part of the decision apparatus; changing it can change the result even when the remaining options stay identical.
One exhausted person can make other people feel lonelier - even without saying much.
Ben Simon and Walker compared participants after normal sleep and a night of total sleep deprivation. Tired participants kept more distance from approaching people; brain scans showed stronger responses in networks monitoring near-space threats and weaker activity in regions involved in understanding others’ intentions. In a separate online test, 1,033 observers rated brief videos of sleep-deprived participants as lonelier and less desirable to collaborate with—and reported feeling lonelier themselves afterward.
Organizations usually treat sleep as an individual productivity variable. It may also be a network variable. Exhaustion can cause withdrawal, which others may read as coldness, prompting them to pull back too. What looks like declining culture or commitment may sometimes be reciprocal distancing initiated by tired nervous systems.
