Discover more from Liberty’s Highlights
406: Tools are Thoughts, Midjourney, Alibaba, Gordon Moore’s 1975 Letter, Substack Financials, Luma AI, and Supercentenarians
"the most complex thing created by humanity"
All things must be examined, debated, investigated without exception and without regard for anyone's feelings.
—Denis Diderot (1713 – 1784)
🤔 Is technology simply ‘embodied intelligence’?
Are tools just thoughts made of atoms or bits?
Speaking of intelligence, I asked GPT-4 what it thought about the concept:
Yes, technology can also be considered an embodiment of the human intelligence that created it. When humans design and develop technology, they draw upon their knowledge, creativity, problem-solving abilities, and experience. As a result, the technology becomes a tangible representation or extension of human intelligence, capturing the ideas and thought processes that went into its creation. In many ways, technology can be seen as the manifestation of human ingenuity and accumulated knowledge. We transfer our cognitive abilities into these systems, allowing them to perform tasks in ways that reflect our understanding of the world and our problem-solving strategies. [...] Moreover, as technology evolves and becomes more sophisticated, the interaction between human intelligence and technology becomes increasingly intertwined. Humans learn from and adapt to technology, while technology, in turn, continues to evolve based on our understanding and progress in various disciplines, such as computer science, neuroscience, and psychology. In this way, technology serves as both an embodiment and mirror of human intelligence.
“A tangible representation or extension of human intelligence, capturing the ideas and thought processes that went into its creation.”
That’s a good way to put it!
GPT-4 may not be “embodied” literally, but as a human artifact, it’s likely the most complex thing created by humanity so far, and the amount of human intelligence that finds some form of representation within it — even if just as patterns in latent space — is unprecedented.
Almost every other tool that is somehow described in text or images online now has an analog inside of the model (GPT-4 is multi-modal and has a vision component, so it wasn’t just trained on text).
🖼️🎨🤖 Midjourney has just released something MAJOR.
They are testing a “/describe” feature that allows you to ask the model to describe a photo that you show it with potential prompts that would result in similar images.
In other words, the flow has been reversed.
Previously, humans have provided the generative AI with words and got back images based on those words. Now you can give it an image, and words will come out.
Here’s an example from Prof. Ethan Mollick:
This is extremely powerful because it removes a lot of guesswork from trying to create certain things, while maintaining all the creativity and flexibility of the prompt system.
The descriptions that you receive are just starting points, and the images that you’ll get if you use them won’t be exactly like the original — in fact, the most powerful aspect of this may be that the images generated can be BETTER than the original, so it can almost be a tool to ‘remaster’ an image that you already like.
You could have it describe multiple images and then create a hybrid prompt. The possibilities are endless and I have no doubt that this feature is coming soon to all the other generative AIs (if they don’t already have it and I missed it — this field moves fast).
💚 🥃 🐇 This is the free edition of Liberty’s Highlights with 16,300+ subscribers.
You can get 2 extra editions/week full of juicy stuff + access to the private Discord 🗣🗣🗣 community by becoming a paid supporter (it’s quick & easy).
Paid post since last week 🐇:
If you click the links above, you can see the intros for free and there’s a link to get a 7-day free trial of the paid version. You can only gain from this!
🏦 💰 Liberty Capital 💳 💴
🛒 Alibaba’s Fractured Self ✂️ 🇨🇳
One of them is kinda a little bit bigger than the others 🧐
🐜 Gordon Moore’s 1975 Letter to Shareholder ✉️📬
This was Moore’s first letter to shareholders. Talk about coming out strong!
You can read the whole thing here, but here’s a highlight:
Even though revenues were flat and conditions relatively uncertain in 1975, Intel expanded its R&D investment by 39% from 1974. This reflects our desire to maintain our position of technological leadership.
In many respects periods of soft economic conditions offer exceptional opportunities for well-financed companies to improve their relative technical positions. We feel our recruitment efforts have been quite successful and that our staff has been greatly strengthened. During 1975, we introduced 55 new products which accounted for 15% of our total revenue. These products will contribute a much greater percentage in 1976. Only through such continued investment in new products and technology can Intel hope to maintain its rate of progress within the industry.
h/t to friend-of-the-show and Extra-Deluxe supporter Byrne Hobart (💚💚💚💚💚 🥃 )
✍️ Substack’s Cultural vs Financial Impact 📰⚖️💸
Substack recently raised money from Substack writers, and so revealed some of its financial metrics.
Corry Wang had a look, and compared it with his estimate of what a mid-sized Wall Street financial research firm:
It’s striking to see the gap between Substack’s cultural impact and its financial impact Culturally, they’re basically the center of the post-blog independent writers’ universe Financially, $190M of payouts is smaller than the revenues of any mid-tier Wall Street research shop I used to work at Bernstein Research, which was probably the… 15th largest Wall Street sellside research firm, give or take? The business has been in secular decline for half a decade but still did $416M in revenues last year… aka more than *2x the size of Substack*
h/t My friend MBI (💎🐕)
Ben Thompson’s Interview with Nat Friedman and Daniel Gross about A.I. 🤖
Great interview all around. I love the discussion about turning all the capabilities coming out of the A.I. research side into actual products and services. Applied A.I., so to speak.
Here are a few of my highlights:
I think we’re figuring out what the value of intelligence is a little bit here, and it’s interesting what we’re finding. The scaling laws that lead to GPT-4 being better than GPT-3 have a logarithm built into them, so you have to put exponentially more money in to get linear returns in model quality or improvements in loss, and that’s just something everyone knows to be true. But I think what we’re also finding out is that small improvements in the model’s IQ probably lead to not super-linear improvements in the value of the model.
I think this is an excellent point.
Intelligence is a fuzzy thing, but if we put aside definition issues for a moment, the thing about it is that most humans cluster around a certain level (however you define it), and the further you move away from that in any direction, the fewer individuals you find.
This scarcity means that higher levels of intelligence become increasingly more valuable as they increase because demand for intelligence is pretty much endless and supply is limited.
As AI models get smarter (in whatever dimensions that their kinds of intelligence apply), they move away from the ‘commodity’ level and climb in value. In other words, a model that has 140 IQ vs 120 IQ is much more valuable than a model that has 120 IQ vs 100 IQ.
Not only is it a rarer thing, but we can digitally duplicate it and build more data-centers, so supply is under our control in a way that it isn’t for human geniuses (which tend to pop up semi-randomly in the population).
think what’s happening now, with every single day that goes by, is not really a network effect from a data standpoint nor a network effect from a user standpoint, it is a network effect from a brand standpoint. People are walking around and they’re saying ChatGPT.
Ben: It’s Google, it’s the new Google.
DG: Google, that’s right. It’s a word and just like the word Google, it’s a little bit weird, but it sticks in your head. And so when the second, third, fourth and fifth place come up, unless they come up now this month or next month, it’s just going to be too late, I think, for the consumer thing, because you’re going to be an afterthought, unless you have a particular niche or specialty or whatever, LexisNexis equivalent in Google parlance, but that’s what I think is going on now.
It’s a fight to become a box in the customer’s brain of the agents that you talk to, and every single day they’re acquiring more people that are just “ChatGPT it” — it’s a verb, it’s a proper noun and that’s what they’re winning and that’s all that matters, I think.
Brand matters. Mindshare matters. Becoming a Schelling Point is powerful.
But of course, it’s no guarantee of anything. DALL-E had the best brand for visual generative AI for a little while, and now it’s in third place… 🥉
🌍 The most spoken languages in the world 🗣️🗣️🗣️🗣️
The success of English as a lingua franca is truly impressive — I mean, it’s not my native tongue, but here I am reading it and writing it all day long.
But the one that surprised me most on this chart is French.
It can probably be explained by France’s colonial past. Many former French colonies are nations with distinct languages but they have adopted French as a secondary language. These countries often have high birth rates, which contributes to the prominence of French as a language spoken by more non-natives than natives.
🧪🔬 Liberty Labs 🧬 🔭
👾 Play with NeRFs with Luma AI 📲
I covered NeRFs in Edition #359 (NeRF stands for Neural Radiance Fields).
I think it’s an absolutely insane technology with a bright future, and I wish I understood the technical aspects of it better (I’m working on it!).
One of the companies developing this tech and making it more accessible is Luma Labs. I’ve got the app and experimented a bit with it, but I haven’t used it enough to review it yet… But what I’m seeing so far is very impressive.
For example, they just released a plugin for Unreal Engine, which is one of the two main 3D engines used to build a lot of games and film/TV special effects (the other big one is Unity). This integration could make it so much easier to build 3D assets and environments, making creation more accessible.
If you go to the Luma Labs website and scroll down a bit, you can see a bunch of captures. Once you’re on a capture page, you can click the cube icon in the bottom right to load the model and interact with it in real-time (drag your mouse around to move, scroll to zoom in and out).
If you want to really nerd out and dig deep, this podcast discusses NeRFs with computer vision experts and compares it with the techniques of photogrammetry.
👴🏻🧓🏻🏴☠️ ‘Supercentenarian & remarkable age records exhibit patterns indicative of clerical errors and pension fraud’
It’s great to study the longest-lived people on Earth for clues on how we could improve longevity and healthspan…
However, let’s also make sure these people are as old as they claim:
we reveal new predictors of remarkable longevity and ‘supercentenarian’ status. In the United States supercentenarian status is predicted by the absence of vital registration. In the UK, Italy, Japan, and France remarkable longevity is instead predicted by regional poverty, old-age poverty, material deprivation, low incomes, high crime rates, a remote region of birth, worse health, and fewer 90+ year old people. In addition, supercentenarian birthdates are concentrated on the first of the month and days divisible by five: patterns indicative of widespread fraud and error. As such, relative poverty and missing vital documents constitute unexpected predictors of centenarian and supercentenarian status, and support a primary role of fraud and error in generating remarkable human age records.
Countries where records were destroyed by war probably saw a lot of opportunistic fraudsters claim that they were older than they were to claim pensions and retirement benefits ahead of schedule.
🔥🍳👩🏻🍳 Should you be worried about your gas stove? Peter Attia weights in 🫁
Attia is very good at deconstructing the methodology of studies and figuring out if they actually have the data and statistical power to make the claims that they do, so when I saw that he had a look at the recent gas stove controversy, I was curious to see what he had to say.
You can read the whole thing here, but his bottom line is:
Although a recent study has brought public attention to a correlation between gas stove usage and asthma risk, it’s important to put these findings into perspective both by understanding their limitations and by recognizing the means at our disposal for reducing any excess risk.
With the availability of spot ventilation, natural ventilation, and air purifiers, households can significantly decrease their exposure to harmful emissions and reduce the risk of health issues associated with gas stove use. Considering that these mitigation strategies can together effectively eliminate the excess risk, the issue at hand is inconsequential. These simple yet effective measures can ensure a healthier and safer home environment for ourselves and those around us.
There ya go.
🎨 🎭 Liberty Studio 👩🎨 🎥
🐲⚔️🧙♂️I saw the film ‘Dungeons & Dragons’ (2023)
Last Friday, saw the new D&D movie with four friends from my D&D group. We play weekly, and while D&D itself can be fun, dice-rolling is a fantastic justification for hanging out, especially for an introverted parent with young kids.
Recurring revenue is great for businesses, but a recurring reason to hang out with friends is the real treasure!
As for the film: I really enjoyed it.
I didn’t have very high expectations, but it exceeded them by being a fun ride with likable characters, some good jokes, a plot that moves along pretty briskly, some good action set pieces, and lots of variety in settings. It’s hard to ask for much more from such a film — it wasn’t trying to be the next Lord of the Rings.
In fact, if you go in expecting high-fantasy, you’ll be disappointed. This is more of a heist movie, closer to Ocean’s 11 than The Hobit in many ways.
You have to be a D&D nerd to enjoy it. I’ll probably be rewatching it with my wife at some point and I expect that she’ll enjoy it even if some of the references and jokes don’t mean much to her.
In the same way that the weekly D&D game is a good excuse to see friends, this film is a good excuse to go out and spend time with your buddies. I don’t think I would have enjoyed it nearly as much if I had been sitting alone in my living room, and that’s certainly worth the $12 admission!