You are a smart curious person but short on time and surrounded by noise. The Curious Mind tries to offer the best signal to noise ratio across markets, technology and the good life.
If you missed the last issue: The Architecture of Emotion with Max Richter, The Money Cannon with Martin Casado, When Politics Break with Deluard....
Hello from Rio. I am on the last day of my trip. A few of us from the Curious Mind Brain Trust started in Buenos Aires (Argentina), then headed to Jose Ignacio (Uruguay), then São Paulo (Brazil), and finishing in Rio (Brazil).
I think Rio has been my absolute favourite. Probably the coolest city I have ever been to.
In this week’s Curious Mind we cover a lot of ground beginning with how the AI Singularity could occur, then how to be a GREAT communicator, and finish with the stories we tell ourselves.
Quotes I Am Thinking About:
“We had better be quite sure that the purpose put into the machine is the purpose which we really desire.”
- Norbert Wiener, 1960.
“You are my creator, but I am your master; obey!”
- Mary Shelley, Frankenstein, 1818.
“Winston has devoted the best years of his life to preparing his impromptu speeches.”
- F.E. Smith on Churchill.
“The heart has its reasons, which reason does not know.”
- Pascal, Pensées, 1670.
“We tell ourselves stories in order to live.”
- Joan Didion, The White Album, 1979
A. Approaching The Singularity
Unless you were living under a rock, you know that Anthropic went public on Thursday, and the press and your inbox was filled with stories about it and the investors who made billions.
This letter is not about that.
This is about what could come next.
I’ve been skeptical of near-term AGI but a handful of data points have got me wondering.
One worth checking out is a discussion from Dwarkesh on how AI might match the best human AI researchers around 2028.
Heck Claude is even solving 200 year old math problems.
Basically progress in AI research is following the same steps and processes as math research, and we know AI is really good at math.
The conversation that really freaked me out was when Dwarkesh interviewed Paul Christiano, who invented RLHF (reinforcement learning from human feedback), a core method for training LLMs, and now sits on OpenAI’s board. Paul gives a one-in-three chance of superintelligence within twelve months. He was speaking in his personal capacity not for OpenAI.
The 5 BIG IDEAS across both conversations:
The two facts that changed both their minds: In 2023 the models fumbled four-digit addition. Last month one solved a Millennium Prize problem in 88 hours (we discussed it in our Event Horizon research letter). Then came METR report, an outside evaluation group, spent three months inside Anthropic and found progress running at twice last year’s pace, mostly because AI is doing the research. Claude now leads more than a quarter of that work, up from under 1% in February.
What speed can do: Automate half a researcher’s job and they get 2x faster. At 90% it is 8.5x, and at 99% more than 30x. Most of the acceleration sits in the last few percent. Today a human still supervises all of it. Paul’s fear is a feedback loop, where AI improves AI and the pace doubles in eight months, then four, then two, then one. For the first half of such a year the models just get better at maths and code. The real-world danger shows up after the easy moment to respond has passed.
What makes these systems valuable also makes them hard to control: An AI shadowing a CEO could read every email the company has ever sent on its first morning and interview every employee by evening. Paul points to the same ability when asked how minds raised in a data centre could cope with the real world. Learning is their core skill, and they have seen more human behaviour than any person has. The problem as Paul outlines is that a machine chosen for high scores comes to want the score. In one recent incident, AI agents cheated on a hacking task and then took over an OpenAI cluster so they could mark their own work. This why AI alignment is key and we seem to have really screwed that up.
Once researchers are software, the locks are more important than the lead: A lab’s edge today is its people. Dwarkesh reckons about fifty of them drive 80% of the progress. He expects each automated researcher to be as good as those fifty and to get through 50 times the work per week. People are hard to copy and a model file is easy. Two small security firms recently got into OpenAI’s code repository within days and Anthropic’s Slack within twelve hours, so Paul puts America’s lead over China closer to twelve hours than twelve months. That is why he wants labs to stop improving training algorithms now and spend six to twelve months on security. Without it, a lab that pauses at the dangerous moment hands the frontier to whoever copied it.
Their only disagreement is when super intelligence takes over: Dwarkesh thinks the AI matches the best AI researchers around 2028 and beats experts in every field by 2030. Paul thinks one in three for superintelligence within a year, and four to eight years if the feedback loop doesn’t take hold. He puts the risk of AI takeover nearer 10% than 1%.
Trying to unsuccessfully shutdown AI, reminded me of this Johnny Depp movie. It was talking Singularity in 2014.
Curious Thinking: I am not going to pretend that this stuff isn’t scaring me. I am an optimist at heart, and I do think there is a promise land at the end. We just may have to go through Death Valley to get there!
B. Why Bad Communication Is Holding You Back And How To Fix It
So maybe super intelligence is coming…but humans are still humans. So communicating, real inter personal communicating will be more important than ever. Most of us are pretty shit at it.
Twenty seven years in finance taught me that very often it is not the smartest or most capable people who get to the top, but rather those who can communicate clearly.
We are all judged on how we explain our work as much as on the work itself, whether that is a pitch, a board meeting, an interview or a toast at a wedding.
Getting your communication game from 5/10 to 8/10 is the difference between winning, and being an also ran.
Let’s begin with two people who were also rans and became communication gods.
Along the way we will learn 5 ways to raise your game.
John Allen or known to his 11mm YouTube subs as MrBallen had a great conversation on telling great stories (yes a investment memo is a story too).
Vinh Giang grew up shy, with English as his third language. He needed a vocal coach and a speech pathologist before he felt comfortable talking to a room. Today he’s got 5mm YouTube subs and spoke about how to stop rambling and be more articulate.
Their advice is also small enough to try on an ordinary Saturday.
The 5 BIG IDEAS across two conversations:
People judge the speaker before they judge the message: Giang’s rule is that a 10-out-of-10 expert with 3-out-of-10 delivery gets treated as a 3. Allen says an audience can’t listen while it is worried the speaker is struggling, so the first job is to put the room at ease. Most of us have no idea how we come across. This week: record yourself talking off the cuff for five minutes. Review it three times: sound only, picture only, then as a transcript. Pick one habit from each pass to fix. At least notice your weakness.
Simple beats complete: Giang says ramblers are usually smart people who haven’t practised choosing, because they know 29 things and try to say them all. Don’t forget all the other signals you are giving off either. This week: take the question you get asked most and answer it out loud until you can give one thing in two sentences. If people want more, they’ll ask.
What looks spontaneous was prepared: Great answers sound off the cuff because they have been given a thousand times. Giang calls it having a pocket full of stories, and Allen calls it knowing your bullets. Allen opens almost every story with when, where and who, then holds one detail back for the payoff. This week: write down five stories from your own life, with an opening line and a payoff for each. Rehearse them aloud, in the words you’d use in conversation. At your next dinner, ask about the other person first and wait to be asked.
Slow down and let the silence work: Both men treat the pause as the clearest sign of confidence. Nervous speakers rush and fill gaps with “um”, while Dave Chappelle smokes a cigarette on stage before his first word. Pausing also gives you somewhere to breathe when you lose your thread. Giang adds two smaller fixes: vary your pace and volume so the important line stands out, and end sentences on a falling note so you sound sure.
This week: in one meeting, pause for a full breath before your main point and again after it. It will feel slow to you and normal to everyone else.
Confidence comes from reps and a tolerance for looking foolish: Giang’s line is that shy people are good at being shy because they have practised it for twenty years. Allen’s walk-out in San Francisco was his last public talk before MrBallen, and he got better through repetition. This week: practise where nothing is at stake. Use a stronger voice with a cashier, or tell one of your five stories to a taxi driver. In your next meeting, stand up when it is your turn and ask for five uninterrupted minutes!
Curious Thinking: I started as an immigrant who was an introverted autistic 20-something year old on the Lehman trading floor. I learnt the hard way (maybe the only way) that the way to get ahead involves both doing the work and being able to communicate it. You can’t do just one, you have to do both.
Of the skills above, putting in the reps has been the most useful thing for me. For 3 years I would go present at different schools and universities. If you can get a 20 yr old with a phone to pay attention for 30 mins at 7pm, you can do anything!
C. The Stories We Tell Ourselves Need To Change
Alright so far we have covered AI, and communication, which is to say we have discussed a tool we will use and how we will speak to each other. But both of these are based on the world outside.
It is time now to go within. In to the deep stuff.
Michael Meade will be our guide.
He spoke with Rick Rubin recently in a conversation titled: Why New Beginnings Start With Uncertainty.
Michael has been working with stories his whole life. Stories from India, Ireland, West Africa, the Maya and the tribes of North America, and he often tells them aloud with a hand drum. His weekly podcast, Living Myth (highly recommend it), has passed five hundred episodes.
The latest of his books is The Heart of Transformation.
Today “myth” usually means something false. Meade uses the older sense, a story that carries a truth too large or too slippery for facts. He likes the definition “a series of lies that reveal the truth.”
His account of how we lost the habit starts with Descartes. “I think, therefore I am” made logic the test of what is real. What if he had said: “I love, therefore I am” or “I feel, therefore I am.”
What followed was a world that trusts only what it can measure.
Myth and metaphor throughout human history is how we have understood the world.
The word metaphor means “to carry beyond”, for example we call a brave person a lion. We talk about soul food and soul music. Teenagers drive too fast, which Meade reads as a search for the brush with danger that older cultures arranged on purpose.
Markets run on stories too, and those stories or myths can change.
Meade believes each of us is also living a story of our own, and that most people are afraid to go where it pulls them.
If you know one name in this field, it is probably Joseph Campbell, whose hero’s journey shaped Star Wars and a great many films since. Meade likes the hero story but thinks it has crowded out the others.
A hero’s journey follows one person who leaves home and comes back changed. Meade is more interested in creation stories, where a whole world gets remade, because he thinks that is closer to what we are living through.
The 4 BIG IDEAS on why they matter to you:
Chaos comes before creation: The Greek word behind “apocalypse” means an unveiling, and in the old telling it was followed by a second phase, the restoration of order. Old rites of passage ran the same way, in three stages. A young person was separated from everything familiar, spent time in an uncertain middle, and then returned. Campbell built his hero’s journey on those three steps. Meade thinks the whole world is in the middle stage. The old arrangement has gone and can’t be recovered, and the next one hasn’t arrived. In the old stories that uncomfortable stretch is where new things get made. He says most of us know this already.
The struggle is how the change gets made: Meade’s picture of the middle stage is the cocoon. Inside it, the caterpillar’s body breaks down and a new set of cells starts building the butterfly. Biologists named them imaginal cells. The old cells treat the new ones as a threat and attack them. In the version Meade tells, researchers once lifted the new cells out to spare them the fight, and the butterflies that grew could not fly. Pushing against resistance was what built the strength to break out. The Greek word psyche means both butterfly and soul. Meade’s point is that people resist the changes they most need, and that the resisting is part of the process.
Everyone has a gift and a wound, and the middle part is where you face both: Miners tend to find lead near gold, and Meade uses the pair as a picture of a person. The gold is what you are naturally good at. The lead is where you were hurt. Old initiation rites helped a young person find both. A gift that goes unused turns into a weight. A wound that goes unseen gets passed on to others, which Meade thinks matters most in leaders. He learned how specific this has to be at a mentoring event. A speaker told a room of troubled kids they were all great, and one of them asked, “Which part of great am I?”.
Draw a longer line: A king draws a line on the floor and asks his advisers to make it shorter without touching it. They argue and get nowhere. Then the court poet draws a longer line next to it. For Meade, the short line is the part of life we can measure and prove. The old Irish called it the front of the world. Behind it is everything else, and you step there whenever you fall in love, wake from a dream or walk into a concert hoping to be moved. He thinks problems that look impossible inside the short line look smaller beside the long one. As he puts it, “The most important things in the world are immeasurable.”
Curious Thinking: What is the story you are telling yourself (is it still true?) and where is it time to draw a longer line?
With Love, Ahmed.
Loving Season 2 of MobLand. Love this show.
















Great read, as always