Building a human-ai collective intelligence platform. @OlenaProtocol
Joined January 2010
- Tweets10.2K
- Following483
- Followers4.9K
- Likes18.7K
And the fact that nothing is obviously broken would be the concerning part if you're invested in this market going up.
Going by current trends in 2 years you could probably just spend nothing but electricity money trying it on your laptop.
The best way to make good predictions is to start with the outside view (the base rate).
Instead of aiding good decision-making subject matter expertise often hinders it b/c being so deeply involved in a subject is a actually barrier to taking the outside view.
17% = LOL
Yes it shows how there's little "crowd wisdom" to be found with LLMs. Humans will have an ongoing role in providing insight for messier questions.
You love to see it.
HEX FOUNDER RICHARD HEART ARRESTED IN ISTANBUL, $600,000 CASH AND ARSENAL SEIZED
Turkish authorities have reportedly arrested Richard Heart, whose real name is Richard James Schueler, in a joint intelligence and police operation in Istanbul.
Heart, founder of HEX, PulseChain and PulseX, was wanted under an Interpol Red Notice issued at Finland’s request.
Police reportedly detained Heart and his wife at a residence in Istanbul’s Beşiktaş district following extensive surveillance.
Authorities seized:
• $600,000 in cash
• 24 wristwatches
• 20 gold bracelets and 7 gold necklaces
• 7 firearms and 1,360 rounds of ammunition
• 12 mobile phones and 6 laptops
• 20 graphics cards and 14 SSDs
• 2 drones
In 2023, the SEC charged Heart with raising over $1 billion through unregistered crypto offerings and allegedly misappropriating at least $12 million in investor funds to purchase luxury goods, including sports cars, expensive watches and a 555-carat black diamond.
Heart was subsequently placed on Interpol’s wanted list in 2024 over separate allegations involving tax fraud and assault.
Both Heart and his wife have been taken into custody for questioning as the investigation continues.
We can be pretty sure there's no way City grew legit income ~£120m in 2018-19 to fill the hole left by ADUG ceasing its hidden subsidies subsequent to the Der Spiegel exposé.
We can also be pretty sure they didn't keep funnelling the cash in via the same route. That game was up.
My guess is the UAE gov is directly subsidising the UAE sponsors for purportedly unrelated activities and the sponsors are passing that value on to City in deals that are still hugely inflated.
An amazing technology but I'm not sure if it'll be an amazing standalone business. Very much doubt it TBH.
Here’s a non-exhaustive list of subject matter experts getting predictions and probabilities wrong in their own fields:
• In a famous medical statistics study, only 18% of physicians and medical students correctly interpreted a positive test result. The most common answer was 95%; the intended answer was about 2%. A replication decades later found only 23% answered correctly.
• Across 13,300+ market forecasts, CFOs supplied ranges intended to contain the outcome 80% of the time. They contained it just 36% of the time.
• Doctors predicting survival for 468 terminally ill patients got within one-third of the actual survival time in only 20% of cases.
• A meta-analysis of 136 studies found statistical methods averaged about 10% greater accuracy than clinical judgment. Experience didn’t eliminate the disadvantage.
• An IMF study found consensus GDP forecasts anticipated only 5 of 153 recessions when made in April of the preceding year.
Moral of the story? People building AI know how to build AI. History gives me plenty of reason to be very sceptical of any predictions they make about this technology however. P(doom) involves *a lot* more uncertainty than the above examples and look at how bad those predictions were.
Basically the exact same thing that's going to happen around AI models.
Football analytics is becoming infrastructure, not a side project.
The @FinancialTimes makes that shift tangible.
Data & analytics have grown to more than 3% of professional football job postings, while clubs such as Brighton, Newcastle and Man Utd have sharply increased hiring.
What interests me is that clubs aren't “using more data”, we passed that point years ago.
The change is organisational.
Recruitment analysts. Data scientists. Analytics engineers. Performance analysts. Data platforms. Models and internal tools built around decision-making.
And that changes where the competitive advantage can come from.
Having access to data is increasingly commoditised (Still expensive for normies). The harder part is building the infrastructure and people that can turn it into better football questions, repeatable workflows and ultimately better decisions.
The moat is moving from data access to data capability.
Frontier labs are going to make a lot more money from complements to their models than the models themselves.
You're only as efficient as your bottleneck allows.
Not sure how many get that if you have a task where AI speeds up 50% of current work by 1000x and 50% by 2x the entire task is sped up by ~4x and not more than that!
There's a System I bias towards thinking it's 500x -- even serious policy people do this -- and this is distorting the AI impacts discourse
The bottleneck doesn't disappear, it moves.
Very cool paper showing that government services are getting flooded because of AI... Echoing @avshah99's earlier findings!
This implies that the bond market will just stop trying to predict what the Fed will do in the future just because the Fed stops speaking about the future.
That's obviously nonsense.
Because the loop matters a lot more than the model does.
And there's a lot more to the loop than agents.md, decisions.md, etc, etc.
If firms are going to fully harness AI, they're going to need to become a lot more efficient at evaluating and recording human decisions too.
Anthropic’s best AI model struggles to attract users as cheaper tools thrive ft.trib.al/42pfSjK
"Overall my view is that AI is structurally a technology that tends to concentrate power, for reasons that have nothing to do with regulation (more to do with the extreme implications of the scaling laws)."
This is only true if the lead time advantage of frontier labs compounds more quickly than the intelligence diffuses.
So far I think the evidence points in the opposite direction. Models are increasingly just one component of larger intelligence systems. Context, specialisation, verification, permissions, feedback loops, etc etc are becoming more important, because raw model capability on its own is not enough to reliably produce useful work.
This is already obvious at the individual level. Most heavy AI users I know are now using 'orchestration wrappers' over multiple harnesses, which in turn command multiple agents using multiple models. The twin goal of the approach is to make more efficient use of the available intelligence and to verify outputs.
I think the same thing is going to happen at the institutional level. We're going to see the emergence of firm-wide harnesses that provide context, permissions, evaluation, memory and orchestration for multiple agents, models and humans. In other words, an operating system for both human and machine intelligence.
This is an area @OlenaProtocol is increasingly focused on as we've spent so much time working on collective human intelligence systems and it's clear this will be a very important part of the jigsaw.
So in sum, the clear direction of travel is that intelligence is becoming more modular and compositional, not less. This is the opposite direction to what Dario is predicting.
1/2 Thanks Gavin for an especially thoughtful exchange. I don't usually spend much time on social media but I wanted to engage here because it really brings out the heart of an important conversation.
First, on regulation, I think that “either concentrate it in the hands of a chosen few companies and politicians via regulation or distribute it widely” is a false choice. I know that there’s a sort of Silicon Valley shorthand where regulation = regulatory capture = concentration of power, but I’ve always found this to be an overly simplified picture of the world. Many people outside this bubble think of regulation as something that constrains corporate power and benefits ordinary people. I don’t necessarily agree with that perspective either, rather I think it’s complicated and really depends on what the “regulation” consists of. But in particular I think that those in the “regulation = regulatory capture = concentration of power” frame often underrate the decentralizing power of objective and fair institutional processes. A crude analogy is that the formal court system can sometimes feel stuffy and elitist, but it does a much better job of defending the rights of vulnerable individuals than the alternative, mob justice. At their best, institutions can vest power in ideas rather than people, and thereby decentralize that power.
This is why Anthropic has always made its policy proposals very carefully. We try very hard to make proposals that disadvantage (slow down) frontier AI companies while *advantaging* smaller competitors. California’s SB53 (which we supported), and even the much-maligned SB 1047 (which we were ambivalent on), completely exempt any company below a certain amount of revenue or model training costs from being covered at all (it was $500M for SB 53, lower for 1047 but we objected to that). More recently the testing process we’ve advocated for at CAISI and the White House involves more rigorous tests for frontier models than off-frontier models — something that differentially advantages challengers. Similarly, the “Pacing the Frontier” letter envisions (or at least Anthropic’s preferred implementation of it envisions) modulating the pace of the very best models while not constraining those who are catching up. This hurts the business interests of the frontier labs and helps challengers, including open-weights!
Overall my view is that AI is *structurally* a technology that tends to concentrate power, for reasons that have nothing to do with regulation (more to do with the extreme implications of the scaling laws). Open-weights do help some with this but are nowhere near a sufficient solution because they simply shift the concentration somewhat to those with the most compute and chips (which are roughly the frontier labs plus maybe hardware providers). By contrast I think the right “rules of the road” can simultaneously (a) address AI’s cyber/bio/alignment risks, (b) institutionally constrain the power of the frontier AI companies, and (c) leave room for open-weights models while also addressing the specific risks that they bring.
BTW I do not think that the events of the last few months have “failed to result in [my] preferred regulatory path”. The approach that the Trump administration is reported to be taking — pre-deployment testing for frontier models, and also testing of open-weights models when they get closer to the frontier — is one that I am very supportive of, though of course I have to see the details to be sure. I am also supportive of Demis Hassabis’ ideas around a FINRA-like entity. This contrasts with six months ago when most of the industry was still pushing for preemption of all state regulation and no apparent federal approach either.