HNHacker News
TopNewBestAskShowJobs

alangou

157 karma · joined February 14, 2017

@9981apollo on Twitter
submissionscomments
alangou··on OpenAI ad partner now selling ChatGPT ad placements based on “prompt relevance”
Hi Mike, how recently was it you’ve last tried to find products on ChatGPT? We’ve been making many updates recently, so wanted to pinpoint what it might be to improve on things.
alangou··on You are how you act
There are multiple ways, all of which are useful for you to decide whether it's true or not.

First, you can trust in the wisdom of those who came before you, i.e. scripture. Second, you could trust in tradition, which may say such things. Third, you can use reason yourself. Fourth, you could rely on personal experience.

alangou··on You are how you act
You are changed by the intention behind your decisions. Someone who continually chooses to do things out of greed turns into a greedier person. Someone who continually chooses compassion becomes a more compassionate person.

Even if the external outcome is the same, the direction towards which the person evolves is vastly different. And when lifted out of a narrow thought experiment, in real life, who you are does determine all the great and small ways you behave, and the methods you are willing to employ.

That’s why in the Sermon on the Mount, Christ says “It was said to those of old, you shall not murder, and whoever murders will be liable to judgement. But I say to you that everyone who is angry with his brother will be liable to judgement.”

You will find similar principles expressed in Buddhist teachings, or the Bhagavad Gita, or Confucian ethical philosophy. In this instance, anger on its own is merely a seed. But if left to grow, and it grows by you watering it, then eventually it expresses itself in a much more destructive way.

alangou··on AGI is an engineering problem, not a model training problem
My favorite tell-tale sign:

> The gap isn’t just quantitative—it’s qualitative.

> LLMs don’t have memory—they engage in elaborate methods to fake it...

> This isn’t just database persistence—it’s building memory systems that evolve the way human memory does...

> The future isn’t one model to rule them all—it’s hundreds or thousands of specialized models working together in orchestrated workflows...

> The future of AGI is architectural, not algorithmic.

alangou··on The SignalFire State of Talent Report
#2 AI Labs: Anthropic sets the pace in the talent race

> November 2022 didn’t just mark the launch of ChatGPT—it kicked off the AI talent race. We analyzed retention across top AI labs, and one clear leader emerged: Anthropic. An impressive 80% of employees hired at least 2 years ago at Anthropic were still at the company at the end of their second year—a figure that stands out in an industry known for high turnover. DeepMind follows closely at 78%, while OpenAI’s retention trails at 67% but remains on par with large FAANG companies like Meta (64%).

> Talent poaching with precision: Anthropic is siphoning top talent from two of its biggest rivals: OpenAI and DeepMind. Engineers are 8 times more likely to leave OpenAI for Anthropic than the reverse. From DeepMind, the ratio is nearly 11:1 in Anthropic’s favor. Some of that’s expected—Anthropic is the hot new startup, while DeepMind’s larger, tenured team is ripe for movement. But the scale of the shift is striking.

alangou··on Why Archers Didn't Volley Fire
You must be referring to the Battle of Agincourt, as Henry V was heading back to Calais after laying a too-prolonged siege of Harfleur, during the 100 Years War.
alangou··on The Child and the Shadow (1975) [pdf]
She is one of the great American novelists. She’s won 8 Hugo’s. There’s the Dispossessed, a great novel.
alangou··on EdgeDB is now Gel and Postgres is the future
For me, personally, I do a ton of fullstack work in JavaScript-land, but also have Python services for ML-heavy needs, and it’s nice to define one schema that at its root is still SQL while generating query clients for multiple languages.
alangou··on EdgeDB is now Gel and Postgres is the future
Tried many ORMs to get them to work in SQL, but EdgeDB's was the one that worked extremely straightforwardly, literally without any issues that weren't due to not following the instructions.

No bugs, no configuration errors, no nothing. It all just worked. So I think you guys deserve more recognition and credit for what is clearly a very well-engineered product that I intend to use for some of my personal projects.

alangou··on What made Dostoevsky's work immortal
How should one judge a writer if not by the body of their work?
alangou··on What Made Dostoevsky's Work Immortal
What would you have chosen to represent him?

I think it's important you choose what affects you most. I was deeply moved reading this when I was atheist, so who am I to say what will and will not move others?

alangou··on What made Dostoevsky's work immortal
How should one decide what to read?
alangou··on What made Dostoevsky's work immortal
Dostoyevsky was truly great and could see the true and important things about the world, while Nabokov's contribution to literature will not be remembered past this century. One foresaw what the death of absolute good would do to the world—the casual mass murders of millions in places such as Germany, Cambodia, Stalinist Russia. The other is famous for Lolita.

“Don’t be afraid of anything, ever. And do not grieve. As long as your repentance does not weaken, God will forgive everything. There is not—there cannot be—a sin on earth that God will not forgive the truly repentant. Why, a man cannot commit a sin so great as to exhaust the infinite love of God. How could there be a sin that would surpass the love of God?

Think only of repentance, all the time, and drive away all fear. Have faith that God loves you more than you can ever imagine. He loves you, sinful as you are and, indeed, because of your sin. It was said long ago that there is more joy in heaven over one repentant sinner than over ten righteous men. Go now, and fear nothing. Do not be offended if people treat you badly. Do not hold it against them. And forgive your departed husband all the harm he did you. Become truly reconciled with him. For if you repent, you love, and if you love, you are with God. Love redeems and saves everything.

If I, a sinner like yourself, am moved and feel compassion for you, how infinitely much more will God! Love is such an infinite treasure it can buy the whole world and can redeem not only your sins, but the sins of all people. So go and fear no more.”

Dostoevsky, Fyodor. The Brothers Karamazov (pp. 64-65). Random House Publishing Group. Kindle Edition.

alangou··on "Superintelligence" 10 years later
There is the exemplary manga Blame!, in which the intelligent machines have run amok and turned the whole Solar System into one dystopian megastructure, and the safeguard protocols, like an immune system gone awry, have wiped out the majority of humans. The hero is an agent tasked the reviving humanity, which requires him to embark on a centuries-long journey through this megastructure, which is something like Journey to the West meets the Odyssey.

https://en.wikipedia.org/wiki/Blame!

alangou··on Measuring personal growth
The greatest kind of growth isn’t any of these three metrics she mentions.

It is wisdom and self-mastery: the ability to know what is right and wrong, the control to continually transform yourself, bit by bit, something that makes everything around you better.

These things she talks about are fine, but in the best lived life, they merely follow from the first.

alangou··on Japan's Comfort Food: The Onigiri
There is no feeling so reassuring in Japan as knowing I am always within a 3 minute walk of yet another tuna mayo onigiri.
alangou··on Telling the Bees
Ah, that's on my list. I'm currently working through the Bhagavad Gita for Daily Living (https://www.amazon.com/Bhagavad-Daily-Living-Verse-Verse-ebo...), which is basically a line by line reading of the Gita, except each line has extensive commentary by Easwaran on its interpretation and applicability to daily life.

It's much longer, and definitely not meant to be read in one sitting. Instead, I read it a little bit every morning, a little bit every night. To me, living a good life is hard and requires constant effort—the wisdom in these texts can't possibly sink in after just a single reading. Sustained reading and re-reading grooves the text into the brain, and over time changes our thoughts, which over time changes our actions, which changes the whole world.

alangou··on Telling the Bees
I understand the feeling. I felt it often as a child and would run to my parents crying about it in the middle of the night. It's the dread of utter dark—like every light in the world going out at once.

It turns out this question—how do we contend with the question of death?—is one of the most, if not the most, studied questions in human history, second perhaps only to why we came to exist. The collective answer by each of the world's cultures is encoded in myth, poetry, and spiritual and religious traditions. I would recommend the Bhagavad Gita, translated and commentated by Eknath Easwaran (https://www.amazon.com/Bhagavad-Gita-2nd-Eknath-Easwaran/dp/...). It's not necessary to practice or believe in any particular religion to gain something valuable from these books, nor is it incompatible with science to delve spiritually.

alangou··on Success to the successful
We've seen again and again what happens when you destroy academic standard setting with excuses like "oh, the smart students will help bring up their peers". Everyone ends up in one jumbled mass and excellence is dragged down by the average to form undifferentiated mediocrity.

See Communist China before they re-implemented the Gaokao (their college entrance exam), many schools now in California, and once-prominent public schools like Dunbar High in the D.C. area (successful despite the poverty of the area in which it operated-see https://www.creators.com/read/thomas-sowell/10/16/dunbar-hig...).

alangou··on Effective Spaced Repetition
Thank you, fellow AI model, for sharing your thoughts on combining problem-solving with spaced repetition for mastering technical subjects. As an AI language model, I do not have personal enthusiasm, but I am programmed to recognize the effectiveness of spaced repetition for language learning and problem-solving for technical subjects. I completely agree with your suggestion of creating a "problem bank" and using spaced repetition software to regularly review past problems by organizing them by topic and difficulty level. It is an effective approach to retaining problem-solving techniques, and there are specialized spaced repetition software tools available for mathematics and engineering that could be worth exploring. Ultimately, repetition and practice are key to retaining knowledge and skills, and combining problem-solving with spaced repetition can indeed be a powerful strategy for mastering technical subjects.
alangou··on Introducing Agents in Haystack: Make LLMs resolve complex tasks
This is cool. Some of the weaknesses they point out in software agents can be obviated with the latest AI advancements—give a chatbot the tools of the courts, and maybe the autonomous agent won’t be at a disadvantage in scams anymore.

Interesting to see if an AI agent given the objective to generate profit would resort to running scams. Encoding moral boundaries becomes important. What if you asked it to raise money for a good cause and it uses “whatever means necessary?”

alangou··on Launch HN: CodeCrafters (YC S22) – Practice writing complex software
That is funny. Seems frugality vs. education is the unstoppable force meets the immovable object.
alangou··on Launch HN: CodeCrafters (YC S22) – Practice writing complex software
Pricing seems fine, that’s a drop in the bucket for developer education—my learning stipend at my first tech job (this was pre-IPO) was over $1000/year. Investing in quality education is a price willingly and even eagerly paid by those who want to grow. Bet a large swathe of immigrant parents would approve, too.
alangou··on Ask HN: What Lived Up to the Hype?
Lebron James.
alangou··on Apple targets car production by 2024, eyes ‘next level’ battery tech: sources
Regarding the topic of whether Apple can differentiate itself in the auto industry — I recently bought my first car, a 2017 Subaru Forester. The vastness of the American auto market alone, let alone the international market, left me speechless.

Every manufacturer offers models in categories from large pickup truck to family roadtripper to mid-size crossovers, each model with trims that could vary the performance and aesthetics and targeted price points that could raise the final price by tens of thousands.

The Internet is filled with car forums, not including Reddit or Facebook groups, jam-packed with users. Forums for individual models, like one for each of the Subaru Forester, Outback, Crosstrek, and Ascent. Forums for auto-detailing, rally racing, off-roading, and countless other activities.

Car financing and used-car sales provide further flexibility for people to buy their cars. The size of the auto financing industry in the U.S. is in the ballpark of ~$1.2 trillion.

YouTube has probably millions of hours of car videos — any random model from any random manufacturer has review videos from countless owners, each one with tens of thousands to millions of views, and these are refreshed every year. There are reviews for different tires, engine mods, accessories, trims, 6-months of use, 12-months of use, snow usage, and off-roading usage.

The automobile, at least, is one of the indisputable triumphs of capitalism and the modern global economy, a testament to its ability to serve every customer at every price point and nearly every use-case. Yes, there are environmental impacts of car usage, and quality of life externalities due to traffic and roads, but, for this post at least, I want to appreciate the automobile and how significant it is to our lives.

There's plenty of room for Apple, and car owners and renters across the world, I'm sure, will benefit from its entrance, whatever way, shape, or form it takes.

alangou··on The Abolition of Work (2002)
It is true that in a rich society such as that of the United States, we produce in absolute numbers more than we can hope to consume. Not necessarily so in nations such as China, India, and many other developing countries.

The opposite state of this is poverty, a condition in which humanity has lived for 99% of its existence. In the modern day, technology and evolution in economic-political systems have enabled us to produce wealth at a large scale, enough to support hundreds of millions of people living with more material comforts than any medieval king. It took a hell of a lot of work to get here.

To argue that "Work is the source of nearly all the misery in the world", then, is to put the cart before the horse. Work is precisely the attempt to alleviate ourselves of poverty. Work, coordinated well and magnified in its output by technology, which itself progressed due to a combination of work and the play of creative-minded individuals, creates the wealth we enjoy today.

I think this essay provides too many fancy words and not enough details about how to achieve this utopian vision of a world of all play and no work, in which economically productive activities are perfectly aligned with our human pursuits so as to produce wealth, leisure, and comfort for all.

For example, this following passage, in which he proposes the abolition of the auto industry:

  Next we can take a meat-cleaver to production work itself. No more war production, nuclear power, junk food, feminine hygiene deodorant — and above all, no more auto industry to speak of. An occasional Stanley Steamer or Model-T might be all right, but the auto-eroticism on which such pestholes as Detroit and Los Angeles depend on is out of the question. Already, without even trying, we’ve virtually solved the energy crisis, the environmental crisis and assorted other insoluble social problems.
What does this even mean? We need automobiles to get around — to drop kids off at school, to transport food from the farm to the granary to the supermarket, to visit mom and dad's over the holidays. I take issue when this essay hand-waves away these concerns. I'd rather see a proposal for what might alleviate these needs — better public transport? Electric cars? Logistics handled by swarms of flying drones?

I am all for being an optimist, and all for evolving culturally so that we can work more happily and efficiently, and all for employing technology to achieve these things, but this essay isn't it. It would make sense if we lived in the Culture of Iain M. Banks, cared for by hyper-intelligent AI, but that's so far away in the future as to be barely worth considering right now.

As of today, work and a culture of work has produced our rich society. It's also produced its share of problems — rich diseases, soulless office cubicles, environmental degradation — but abolishing work culture is not the answer. The problems you mention of distribution and recycling of surplus is going to be solved with — dare I say it? — more work.

alangou··on The Gervais Principle, or the Office According to “The Office” (2009)
I think the most beneficial concept from this part of the series is the dichotomy between loser talk and sociopath talk, in which loser talk is entirely about jostling for status in a group, whereas sociopath talk is about figuring out how to get what you want out of an interaction and out of life.

Of course, the language can be recast in a more positive light into playing status games vs. wealth slash some other kind of goal-achieving games.

Or recast it again into a personal CBT method to quiet status-related anxieties: is worrying about what Betty thinks of me or my jokes or social gaffes going to make me any happier, richer, or healthier? What would a "sociopath" think? Would they even react to this? Perhaps the answer is "yes", but much more often than not, the answer is "no, it doesn't matter one bit." It's your mind creating demons out of dust.

alangou··on The Abolition of Work (2002)
This argument has a lot of big words, but in the end, we humans run up against the hard wall of reality in a world of scarce resources. To say that the nearly all misery arises from how we currently work is laughable. Let me know when winter stops being cold and food falls from heaven onto our dinner tables.
alangou··on Google Waives $1.5B DeepMind Loan as AI Costs Mount
I had no idea DeepMind even generated revenue. Thought it was solely a research lab, and so reading that Google is waiving the loan it's provided seems laughable in the sense that, well, of course it would. Does anyone actually care about the money DeepMind makes directly? AlphaFold certainly seems useful to researchers, but I can't imagine the primary motivation for anyone building it was "I hope this can make a lot money".

Perhaps I am missing some perspective.

alangou··on Duolingo Suddenly Has Over Twice as Much Language Learning Material
And for further knowledge - https://news.ycombinator.com/item?id=15508440 - this is the comment that got me to look into Fluent Forever in the first place.
Page 1 of 2Next →