FWIW, you're not wrong. As I said in my other comment, I don't really know what other models are up to but Claude is definitely demonstrating a deep level of understanding, and I learn things from it all the time.
Sometimes I'm like...is everybody else using the same thing that I am? This is insane.
They definitely let it churn until it figures something out, it's not uncommon for it to take 15-20 minutes to answer, is that normal? They just have "relentlessness" maxed out and some kind of second-order structure to orchestrate a bunch of agents into one unit of work.
It's obvious that what we're getting are summaries. Sometimes I think I even see vestiges of the summarizing model instructions. But, that being the case, it's bizarre why Gemini's summarizer acts like a toddler, and Claude's actually lets me learn and, more importantly, stop it at its tracks when I see it's going down the wrong path.
On the other end of the spectrum, if you toy around with the open-weights models, you see the insanity that is going on in their output. I can totally believe in a scenario where claude is basically doing that and then summarizing it into something coherent, getting rid of the 10 opposing viewpoints, "aha!" and "Hmmm, but wait!", but it does it elegantly.
I've tried constantly to use other models (and I do), but Claude spoils me to no end. I keep thinking I must be holding everything wrong.
I mean, it does though. The sequence of thoughts that Claude goes through in response to my queries blew me away the first time I used it, and it continues to impress. Just the act of putting those sentences together demonstrates such a deep understanding of what I'm saying. If it's just an approximation, whatever it's actually doing is even better.
I haven't even really tried to use other models in the same way, and this thread makes me feel like I don't need to.
Hard to get too worked up about that when the US govt has allocated about 500 billion to Lockheed Martin for a jet fighter. I'd honestly put the jet and the docs at about the same expected value ("what exactly is this doing for our people?").
To me - and I would guess this is why others don't like it - this is the distillation of AI slop.
People get the value of AI all wrong. It's not supposed to be a one-and-done button. This mentality is why people think AI will make us dumber.
AI is best as a tool to probe the extents of your own understanding of a problem and identify gaps and vagaries. You should interact with it with a full keyboard mentality like you would a fellow human. In this way you will actually learn many things from it that strengthen your own understanding.
And yeah, on top of that, it will "do all the work" super fast. It's the ultimate pair programmer.
I dunno. I'm hardly some "run 20 agents overnight" kind of guy and I'm not an ASI guy, but this sentiment sort of reads like a monk who makes a living copying manuscripts, dictating how the new printing press technology should be used.
We are about three or four years in to a technology that has the potential to be world altering! This is absolutely the time to experiment with form factors and techniques, and challenge our own assumptions of what it is we actually do.
And even if you reject all of that, a utility like this one isn't even incompatible with your vision! Couldn't you program a key to say "explain this concept in clearer terms"? However you work with the tool, this gives you ways to use it better. That...seems really nice!
That's fascinating to me. I'm hitting usage limits on Claude Max and I think we've basically reached the singularity. I was moved to write an essay about it.
I'm fully there with ya but I don't understand how a device with such limited expressive power is bringing anything to the table. If you're just using voice, just use voice? Sure, in some sense, it's always cool to try ideas. But IMO anything new would try to improve on subtlety and richness of communication.
Speech can do it all but eventually fatigue sets in if you do try to do it all with speech. Like RSI from too much typing but much faster. Best to use speech for the occasional long prose and steering, and something else for the yes/no/accept/decline options so the throat gets reasonable breaks.
I think it depends a lot on the nature of the city, its population, and the infrastructure choices they've made. People that have lived different experiences may be talking past each other. Most people don't have extensive experience in a wide swath of cities.
In NYC, it's an absolute melee and all modes are in the way of each other. Places like Amsterdam have a proud biking heritage. In a more suburban US city, it's pretty much all car on car, in a lot of situations being a pedestrian is crazy and unsafe, but bikes can coexist on the large open roads fairly well.
IMO prioritizing cars has had overall disastrous and dehumanizing effects. I think it's right for places to be pushing safer, more accessible, and less energy-intensive modes of transport, but there's a huge amount of inertia to fight at this point.
So what exactly is your beef with urbanism? Would you prefer if we just focused on making car traffic as fast as possible and pushed aside other methods of transportation? Do you interpret these results as evidence that attempting otherwise is folly?
They "worked" in some places in Europe because they made traffic so shitty that more people were forced to use bikes. In turn, this provided driver education, by showing them how badly average drivers behave.
That sounds like a pretty big win to me? Yes these programs are designed to make other modes of transport more attractive. What do you make of all the data that shows that adding lanes to a highway doesn't decrease traffic?
It's the root cause of all the current political problems in the US and Europe. It's a kind of pollution that offloads externalities (high living costs) from companies onto the population. Dense cities allow companies to more easily access larger pools of workers, without having to care about their long-term benefits.
> Would you prefer if we just focused on making car traffic as fast as possible and pushed aside other methods of transportation?
Yes. I would go as far as ripping out all the local transit (eventually). Self-driving cars have solved the only real problem with cars: people who can't or shouldn't drive (children, elderly, people with disabilities, etc.)
> Do you interpret these results as evidence that attempting otherwise is folly?
Yes. Quite simply: look at the stated goals of programs like Vision Zero. They are social engineering programs that simply use saving human lives as a justification to push their agenda. I hate that.
>Yes. I would go as far as ripping out all the local transit (eventually). Self-driving cars have solved the only real problem with cars: people who can't or shouldn't drive (children, elderly, people with disabilities, etc.)
One other problem with cars is that not everybody can afford one, or really should need to have one in a big city. Even if Waymo were a free or cheap public service, it's still far less efficient and more disruptive than a robust public transit system. When dense cities are walkable and bikable, less people need cars for everything. That's hard to achieve with cars at 40mph everywhere. There's a place for cars but we've only created the problem by having them zoom around every inch of a city at the expense of everything else.
Also, tax the companies, easy. In America we've been manipulated to loathe the ideas of taxes. Taxes are what pays for things that are good for everyone.
> One other problem with cars is that not everybody can afford one
That's incorrect. A typical bus/subway ride costs around $20, about the same amount as a car ride. We just heavily subsidize them, typical farebox recovery rate is around 15%, and that's not considering capital expenses.
Once shared self-driving cars become more common, we can just retarget the subsidies.
> Even if Waymo were a free or cheap public service, it's still far less efficient and more disruptive than a robust public transit system.
Self-driving cars with mild car sharing (2-4 people per car) will be more efficient than transit in almost all scenarios.
> When dense cities are walkable and bikable, less people need cars for everything.
Sigh. My point is that dense cities are now unliveable for the average person.
> or really should need to have one in a big city.
There shouldn't _be_ such cities. Giant metropolises are a failed experiment.
> Also, tax the companies, easy. In America we've been manipulated to loathe the ideas of taxes. Taxes are what pays for things that are good for everyone.
Yes. One good option is to tax the dense office space so that companies are incentivized to do remote work and/or offices in smaller cities.
There shouldn't _be_ such cities. Giant metropolises are a failed experiment.
I see what you're getting at now, but they're as popular as ever (arguably more so) despite the legacy flaws. With what we know and the tech we have now, the cities of the future could wipe the floor with anything we have today, if we had the gumption to do it right.
That's just how the human psychology works. If you can't avoid something, you'll justify that it's fine and you actually are enjoying it.
And the problem is that an average person _can_ _not_ avoid living in large cities. That's where all the good jobs are. My favorite example here is Tokyo: the housing market there is now completely unaffordable. In a country with _declining_ population.
This is not healthy. And this is _the_ reason for the recent unrest in the Western world. People are seeking easy answers: it's the fault of "corporations", migrants, NIMBYs, foreign investors, private equity, AirBnB hosts, etc.
> the cities of the future could wipe the floor with anything we have today
By not having any residents?
Imagine that you have an option to live in a single-family area with nice parks and plenty of personal space versus a closet-sized apartment on floor 11. What would you choose?
That's just how the human psychology works. If you can't avoid something, you'll justify that it's fine and you actually are enjoying it.
White flight and urban decay already happened though. Why are people moving back?
I believe the reason for the unrest is that technology and the attendant wealth consolidation have greatly outpaced our social systems and constructs. Life should be easier for people. We have the resources. We just don't have the collective will to make them work for the greater good. But yeah it's easier to point the finger at the bogeyman of the week.
You're referring to ideas I would consider to be in the tradition of the Garden City Movement. Certainly many people will always prefer the pace and the space of suburban life. But spend some time in the Eixample area of Barcelona if you get a chance. It doesn't have to be as crazy and inhumane as Manhattan or Tokyo. Some people figured out healthy city living 150 years ago and for some reason (cough, capitalism) the world largely ignored that information.
A reboot of those ideas with pre-planned modern infrastructure would crush. Move most high speed traffic underground. Make the ground level for people. Imagine an underground Waymo-Metro network that went in before the city above it.
> White flight and urban decay already happened though. Why are people moving back?
Jobs.
"White flight" is also a US phenomenon, yet over-urbanization is happening everywhere. Look at Japan for an example. And it's a mono-national country without any significant internal racism or segregation.
Tokyo is also a nice counter-example for many proposals. It's building new housing as fast as possible (whole neighborhoods are construction zones!), it has a land value tax, it allows microapartments, it's building transit all the time. Yet nothing helps. Housing prices keep rising.
> I believe the reason for the unrest is that technology and the attendant wealth consolidation have greatly outpaced our social systems and constructs.
And over-urbanization is one of the reasons this became possible (not the only one, but definitely a contributor). An average company in a small town can't afford to treat people there like shit. They simply won't have any workforce or customers after a while. A company in a large city always has access to a pool of candidates.
> But spend some time in the Eixample area of Barcelona if you get a chance.
I grew up in Russia and Ukraine, and I also lived in Europe (Germany and the Netherlands). I got my driving license at 27 years and my first car at around 30 years of age.
And Spain has all the same problems as other countries: large cities are unaffordable, and small cities are dying. It's more relaxed in general, but the trend is unmistakable.
> Life should be easier for people.
Yes, and the only way to do that is to look at the root causes.
"White flight" was only a response to your assertion about human psychology. My point is that I think more people find urban life desirable and enjoyable than you give credit for. I don't care much for NYC, but it seems like every time I go to Europe there's those big ugly buildings on the wall. A restauranteur in Vienna once took us back to show us his memorabilia room with twin towers poster and all.
I think "make urban life better" has much higher potential than "convince people they don't want to live in big cities." I see the root causes as social and political more than technical. There are certainly advantages and efficiencies to be had with a dense population.
Unaffordablity due to wealth inequality and concentration is the major social dilemma of our time and I can't really connect it to people pushing for more bike lanes and park spaces.
Look up Robert Moses (NYC). Or Georges-Eugène Haussmann (Paris), or Ildefons Cerdá (Barcelona). These are general urban planners not strictly concerned with biking, but single individuals can have incredibly extensive effects on huge populations, that may last for centuries (for better or for worse).
In the US, not a single person, but the General Motors/Mack Truck/Firestone Tires/Standard Oil cabal exerted significant influence on shaping how our cities are built. Literally buying up streetcar companies across the nation to put them out of business, among other tactics.
You can't introduce bikes without increasing the rule complexity and worsening the experience for everyone else.
Do you find two-way car traffic similarly inherently problematic? Left turns across oncoming traffic? I generally agree with you but it feels a little odd to rail against bikers in this way when car on car traffic is far more dangerous. But in car culture (which I am a card-carrying member of), we don't mind that!
It's true though. It's not like math or software where you can do it isolation and work things out logically "on paper." I don't think the solution will be "put a bunch of agents on it and let them figure it out." The tools to model and observe what's going on in full detail don't exist. Far from it. You can't really optimize the trial and error feedback loop time.
I think it will look more like Claude just helping every expert along the way do their job better. Analyzing and synthesizing data to make better experimental suggestions and design new equipment (or totally new experiments based on unconceived-of apparatuses?). Managing the flow of information and building bespoke knowledge bases and tools. The more it is integrated into all the systems and researchers involved the faster it can put things together.
Although, on second thought, maybe there even is a near future where it takes our models and understanding and just galaxybrains a better model. Has anyone tried feeding it some collection of datasets and saying "can you make a more detailed model of a cell?"
> You can't really optimize the trial and error feedback loop time.
Isn't this exactly what startups in this area are doing - automated experiments connected to AI. Of course at some stage things will "graduate" to human review but it still seems to have a lot of potential.
Even with automated experiments, the wet chemistry is slow. The kind of answer you might calculate in a second for a math problem could take many hours for the chemical reaction to occur.
Biology grad students can spend 40+ hours per week in the lab, on top of all their course work. It's not so bad though, since there's plenty of time to grade papers or work on your thesis while waiting for reactions to occur.
Yeah this, I used to sit in the lab with my girlfriend (now wife) when she was doing wet work for her pharmacology doctorate. It was slooooow. Like slower than J2EE. I could smoke a pork shoulder faster.
Or in other words: the bulk of the work is already automated! The conditions have been set up and the reactants are doing the work of reacting. There’s nothing for the experimenter to do but wait.
The reaction can’t be sped up, its rate is wholly constrained by the nature of the compounds involved and the physical parameters of the experiment, which may be very narrow to avoid unwanted products or even thermal runaway which could cause a total failure of the experiment or even a dangerous release of poisonous or explosive materials.
tldr; No sane drug professionals would have designed a relatively simple compound (144 Atomic Mass) as a Multiple Sclerosis drug
I think human review may screw things up. If you look at dimethyl fumarate[1], no card carrying medicinal chemist would have taken this serious. No pharma exec would have risked 100s of millions of dollars for clinical research trials.
- 1950s people notices that a wood polish cleared psoriasis. Topical formulation worked.
- 1990s an oral formulation was created.
- 2000s people noticed people treated for psoriasis that had MS, had fewer MS symptoms while taking this molecule.
- 2017 dimethyl fumerate is approved for MS.
[1] https://en.wikipedia.org/wiki/Dimethyl_fumarate
>The precise mechanism of action of dimethyl fumarate is not clear.
And here is the real problem. Instead of burning gigawatthours of power with GPUs, if Anthropic really wanted to help bioscience, they should just be throwing their $2 trillion at making it possible to live off of studying things like this.
We aren't in a place where a magic "Intelligence" could just look at the data and solve all our problems because we have never funded the collection of that data in the first place
We spent a lot of money on the Human Genome project and got large dividends from it but seemingly never reinvested! There's so many clear and straightforward questions we should be answering in biology that we just aren't, because nobody wants to pay for it.
> You can't really optimize the trial and error feedback loop time.
They can put it onto robots / automatic labs, so this step is easily
solvable. Humans in a modern lab are largely unnecessary. Just look
at China.
The problem here is that Anthropic claims to have solved this all. It
is time that they actually prove their claim. Right now it reads like
an advertisement to inflate the stock of Anthropic.
> They can put it onto robots / automatic labs, so this step is easily solvable. Humans in a modern lab are largely unnecessary. Just look at China.
You can automate the setup and mechanical process. You could increase the physical capacity for parallel tests. You could increase the rate of acquiring raw materials.
But how much you cut down the time depends on how much of the trial-and-error feedback loop depends on feedback and iteration vs cast a wider initial net.
(One of the interesting things to me about the dominant software development agent pattern across all of Claude, Codex, and Cursor is that the approach is still extremely local and iterative, human-mimicking, rather than big-brain-one-shotting a novel or perfectly-fit-for-purpose software design. The performance of the agents on the initial "jump to conclusions" step does not appear superhuman in the same way that the speed-of-iteration is clearly superhuman.)
That is utterly untrue. It stems from the notion that biology only needs big data for us to solve all its mysteries.
Beyond that, while automated labs for cell biology are plausible and active, and the same fit organoid biology, the same really isn’t true at organism level. Labs already are artificial environments that make organisms have odd phenotypes, and odd interactions which bias the data away from what you’d find in the natural world.
You add in robotic manipulation, and you add a new confound.
Drones in the wild are an obvious counterbalance to this, though even in that scenario, the drone can be a confound unless well hidden.
In the end though, this kind of systemic breakdown of biology, while much needed raw material to improve our theories, won’t solve the whole of biology. Human hands and human minds will be needed for a good long while.
Society? Why are you less crazy about Musk properties than any of the rest of the S&P? Does that suggest that a system that allows such concentrations of wealth and control in so few individuals might be flawed?
IMO, let the unicorns leave. Europe figured it out. Get the vast majority of the benefits of their contributions without being sucked dry and subjected to their whims in whatever other domains they fancy.
Sometimes I'm like...is everybody else using the same thing that I am? This is insane.
They definitely let it churn until it figures something out, it's not uncommon for it to take 15-20 minutes to answer, is that normal? They just have "relentlessness" maxed out and some kind of second-order structure to orchestrate a bunch of agents into one unit of work.
reply