
ai auglab @mit organized India's first AI film festival yc s23 intel isef
sf
Joined August 2014
- Tweets3.5K
- Following1.9K
- Followers954
- Likes25.4K
i gave a fruit fly access to internet and let it doomscroll.
Average Reel watch time:
Humans: 14 seconds
Fruit fly: 8.5 seconds
Internet is going to get very fun when animals come online!
For the first time, scientists have mapped the complete brain and central nervous system of an adult male fruit fly — a key model organism in science. 🪰
Working alongside HHMI Janelia Research Campus and the scientific community, @GoogleResearch scientists and researchers used AI to combine millions of 2D images into 3D neural shapes, reconstructing a record-breaking 166,000+ neurons. This foundational map of the adult male fruit fly brain can help accelerate our understanding of the brain, and is a major milestone in neuroscience.
RANA G retweeted
ah cool, math has fallen you say? how about we do traffic, air pollution, factory farming, and disease next? and preserve every living species that will go extinct in the next decade? interspecies communication, interplanetary travel? let’s get existential then, shall we?
RANA G retweeted
Hot take: A country cannot become a manufacturing superpower if every young person thinks working in a factory is something they've failed at in life.
what is a moat in the era of ai?
A business takes a set of inputs, generates an output that likely sells for more than the cost of inputs.
If the output becomes copyable in 10 years, the business has a moat for 10 years.
If it becomes copyable in 5 years, the business has a moat for 5 years.
1 year … 1 year.
1 month … 1 month.
Copyable in 1 day, the moat lasted for 1 day.
Now say, the business takes its previous output, along with a set of inputs, generates another output that sells for more than the cost of creating previous output.
If the final output is just as copyable, then such a business would not have any moat.
If the final output needs the last output to build on top of (scientific research, tech tree, data), the copier has to work back every loop. In mathematics, we call this phenomenon computational irreducibility.
The moat of such a business is the number of irreducible loops it will take for a competitor to catch up.
It follows from that, in such market conditions
It is better for people to stay in stealth much longer, and build irreducible companies
- scientific research, because discoveries are path-dependent until published
- manufacturing, because yield is learned by running the line
- biology, because clock is set by the literal interactome
- safety, where regulation approvals come with accumulated operation
etc
But hey not everything needs to be profitable to be built, gotta take a crack at it!
Replying to @adamcohenhillel
the problem with working on the interface layer is that any solution you come up with propagates to all your competitors ~instantaneously at zero cost to them, as soon as it's validated
2
171
This is a wonderful distinction, and a net positive for the science I believe. When proofs and observations are rewarded separately from their explanations, both will become more frequent and richer simply because more people will compete on each.
A welcome side effect will be on education. k-12 lags decades behind the frontier in the hard sciences because we discover new ideas much faster than we discover best ways to teach them to children. If explanation becomes a first-class contribution in its own right, that competition will significantly lower the barrier to entry to science, and thus kicking off a flywheel for the inchoate society of geniuses.
Therein lies the bitter lesson for science. Any human, taken apart, is a combination of an observer, a data collector, a prover, a theory-builder. Instead of identifying as any or all of these, and as a result crashing out every time a machine does something better, identify yourself as, for now, a participant, but in general a recipient of the process of truth-seeking. We mustn’t let hubris come in the way.
If Newton had only observed the apple fall and asked why, that question alone would have led some other chap to gravity. And Eintein, the theory-builder he was, would likely have developed general relativity all the same.
Replying to @_AashishReddy
If Levent is allowed a gap between publication and explanation -- that courtesy must be applied to everyone.
RANA G retweeted
Parth, India disappoints both pessimists and optimists.
But your dharma is to be an eternal optimist.
RANA G retweeted
BREAKING: Robert Pattinson will play Claude in an upcoming biopic on the Riemann Hypothesis.
RANA G retweeted
A simple test for what the best and brightest should work on:
Does civilization need this?
Would this exist without you?
AI marketing tools, online gambling, email for agents. Either nobody needs it or someone else will build it anyway.
Aim higher!
Hey @Google, it’s retarded to take someone’s suicide note down.
I don’t know who this person was, but to exercise authority over someone’s last words like this is callous and deeply unconscionable.
3
156
RANA G retweeted
India is the land where western political terminologies come to die.
"Liberal progressives" supported a party that was perfectly fine with trans people begging.
"Fascists" design policies to give destigmatize transgenders and help them earn a dignified living.
#WATCH | Kolkata: West Bengal minister Agnimitra Paul says, "We are launching a new initiative: we are forming a brigade consisting of our transgender friends, which we have named the 'Brihannala Brigade.' We intend to engage them in two ways, with the primary goal of providing them with a dignified livelihood. It is the wish of the Chief Minister that they no longer have to resort to begging at traffic signals; instead, we want to offer them a respectable alternative means of earning a living. To this end, we have decided that those among them who are interested will receive training from the police and be deployed at traffic signals. Their role will be to sensitize the public...we are involving another group of transgender individuals in the collection of outstanding taxes. Given our current staff shortages, we often have to rely solely on sending out notices; these individuals will now assist in this process. They will work alongside our inspectors, and we will provide them with training—including grooming—on how to interact with the public and how to approach them regarding tax payments. They, too, will receive a stipend for their work."
RANA G retweeted
"The men that conquered this land slept in mud, and their grandkids complained about the temperature of the wine"
-Julius Caesar
Readers added context they thought people might want to know
This quote is not from Julius Caesar. It appears nowhere in his writings or ancient biographies such as Plutarch's Life of Caesar.
en.wikiquote.org/wiki/Gaius_Jul…
graciousquotes.com/julius-caesar/
perseus.tufts.edu/hopper/text?do…
RANA G retweeted
random thoughts on the Anthropic IPO:
> Anthropic is a uniquely evil company, and I think everyone would be better off long-term if it imploded. Hence it feels weird so much wealth is being created by the ipo
> as an emerging manager (and to be a hypocrite as well), I hope the stock does not dump before the lock up ends and LPs can sell their shares. The resulting liquidity helps keep people like me in business (and would be very very bad in the reverse case, despite my fund having 0 exposure to Ant)
> The market cap makes no sense, most private market investors at least assume there is very serious business model and debt default risk in the business, yet the same circle has been hyping this stock up to “peasant” retail investors as a must-buy insurance against the permanent underclass. This translates to an extraordinary wealth transfer to private market investors, who will make a killing, from retail investors who I suspect will be lucky to make a modest return from holding the stock (if not get totally wrecked).
> the previous point creates a bad precedent (obviously self-enrichment by dumping bags on retail is not great), among other things because it will reinforce the idea that we need to continue to overcook businesses in the private market (“because it worked for Anthropic”) despite the empirical fact that this is a great way to kill companies (and subsequently stifle innovation)


























