Dennis Hackethal’s Blog
My blog about philosophy, coding, and anything else that interests me.
Tweets
An archive of my tweets and retweets through . They may be formatted slightly differently than on Twitter. API access has since gotten prohibitively expensive – I don't know whether or when I'll be able to update this archive.
But in case I will, you can subscribe via RSS – without a Twitter account. Rationale
I don't think he claimed that civilization selected for good explainers. And I didn't write my book "just on that."
Induction is impossible (see Hume). We've known this for ~250 years, but almost everyone ignores this!
David Deutsch's "Beginning of Infinity" and his article "Creative Blocks: How Close Are We to Creating Artificial Intelligence?"
There's also my "A Window on Intelligence" if you want to count that.
Me saying "widespread misconception" already implied that my definition was not common.
The other, common approaches to AGI have been refuted. So my definition of AGI isn't just "idiosyncratic." I can supply the necessary sources if interested.
It makes all the difference because AGI is the project of explaining how the human mind works. That's an epistemological question.
It having to do with "learning tasks" is a widespread misconception.
I don't see how any of that tells us anything about how the human mind works.
I suggest reading it again. Maybe more than once. Especially chapters 4, 5, and 7.
Judging by your book's outline on Gumroad, it may help you correct some errors so you don't head down blind alleys.
I wasn't talking about the brain, I was talking about the mind.
And, because intelligence is a universal phenomenon—recall Deutsch's concept of the universal explainer—any simulation of it is qualitatively the same (modulo implementation details).
Your prediction is self-contradictory because a "discovery of AGI" is an explanation of how the mind works.
This relates to what you wrote here: twitter.com/IntuitMachine/…
No need to refer to any "properties of living things" because intelligence is an emergent phenomenon. We know this from computational universality.
Cool. I also see you mentioning "Jumps to Universality" and "Constructor Theory." So... you've read his book, just not chapter 5?
Having glanced at the outline you present on Gumroad, you may benefit from reading aeon.co/essays/how-clo… before you publish your book.
E.g. a boiling pot of water. We can explain what happens there without calculating each position of each molecule at every second. The behavior of the pot is simpler on a higher level of emergence and is best explained on that higher level.
@SpaceX @NASA @Astro_Doug @AstroBehnken @Commercial_Crew
Very nice. Onward!
A phenomenon is emergent when it is best explained without referring to its lower-level components. Compare David Deutsch’s “The Beginning of Infinity” chapter 5.
No human can yet explain how the mind works. But one day, we shall. Problems are soluble.
“A woman had gone insane from excessive riding of the bicycle.” 😂
Though cool, GPT-3 is not a step toward AGI.
AGI is the project of explaining how the human mind works, and then implementing it on a computer.
I'm not aware of any insight GPT-3 gives us into how the mind works, nor is OpenAI after that (sadly).
@IAM__Network
By definition, AGI and human intelligence are the same.
The project of AGI is to explain how the human mind works, and then implement it on a computer.
Since nature achieved human minds somehow, and since computers are universal, AGI must be possible.
More generally: any good theory of AGI explains how the human mind works.
Pretty far, I'm afraid. We need to explain how the human mind works. Only then can we build AGI. That's what most researchers fail to realize.
@pmoinier @JulienSLauret @TDataScience
GPT-3 is very cool, but I'm afraid it's not AGI, because it doesn't explain how the mind works.
I'm a minute in and he mentions induction, which refers to an impossible process of knowledge creation. He won't build AGI if he goes down that road.
The primary question in building AGI is: how does the mind work?
Yes. The primary task in building AGI is understanding the mind.
@tjaulow
And yours truly is working to create a course for beginners, so stay tuned for that!
@tjaulow
You bet. I started with a book called “HTML For Dummies.” Though not a fully-fledged programming language, HTML will feel like one, and the visual element will help you correct errors quickly. After that I learned CSS and JavaScript.
I think what’s good or bad is objective.
Regardless—yes, we can’t predict its preferences and goals either way.
Not at all! I’ve seen people successfully switch careers to coding in their forties and fifties (eg from law). I think people at any age and any career stage can do it.
@GoodNPlenty333 @nburn42 @Plinz @aifdn
I had already given a metric: that of whether moral knowledge solves a problem, and how many. But no there’s no metric to just solve all moral problems.
This is just another surface issue though. Like I said, we will keep having disagreements if we don’t agree on epistemology.
@GoodNPlenty333 @nburn42 @Plinz @aifdn
Not much hinges on this as we disagree on a thousand surface issues but maybe just one underlying issue: epistemology. Not much point in discussing unless we agree on epistemology.
We can then ask: but why were those genes better able to spread? Which seems to me analogous to what you're looking for (correct me if I'm wrong).
Well, it's a bit like asking: why do male peacocks have such elaborate tails?
One answer is: because female peacocks like them. The deeper answer is: because genes that happened to code for slightly more elaborate tails spread better.
I think it's the same with memes.
*watch on a heath ;-) But yes! A great thought experiment.
@GoodNPlenty333 @nburn42 @Plinz @aifdn
Guys don't get laid by fighting other guys, either.
@GoodNPlenty333 @nburn42 @Plinz @aifdn
I'm happy when I play video games. Video games don't get me laid nor do they make me rich.
@nburn42 @Plinz @GoodNPlenty333 @aifdn
Again, mental phenomena are emergent. We should explain them without referring to any underlying hardware. We know from computational universality that you can run the same phenomena as software on a laptop, even though laptops are not made of neurons.
@nburn42 @Plinz @GoodNPlenty333 @aifdn
When you speak of causes, you're giving an explanation. And an explanation along the lines of "it's all just neurons firing in some way" applies to any mental phenomenon (happiness, sadness, curiosity, depression, etc) and so it's a bad explanation: it may as well apply to none
@nburn42 @Plinz @GoodNPlenty333 @aifdn
Happiness is an emergent phenomenon. As such, it should be explained independently of the underlying hardware (brains/neurons).
@DoqxaScott @Plinz @GoodNPlenty333 @aifdn
I'm basically arguing against historicism applied to the future.
And the further we try to look into the future, the less reliable (and meaningful) our predictions become.
That's not to say that making predictions isn't useful!
Yes—I just recently discovered William Paley's "Natural Theology" to be a treasure chest of good ideas (albeit mistaken conclusions!).
Can be read for free here: google.com/books/edition/…
@Plinz @nburn42 @GoodNPlenty333 @aifdn
"And solving problems itself depends on knowing how; so, external factors aside, unhappiness is caused by not knowing how."
@Plinz @nburn42 @GoodNPlenty333 @aifdn
David Deutsch offers an interesting conjecture in his book "The Beginning of Infinity": "Happiness is a state of continually solving one's problems [...] “Unhappiness is caused by being chronically baulked in one’s attempts to do that."
The primary reason in any case we can know from meme theory: those inconsistencies above exist simply because their respective words managed to spread.
Ah yes, my mistake.
The problem remains, however, as oftentimes the opposite argument is made: that simplifications are made to words people use often (eg so that spelling gets easier/shorter etc).
A solution to a moral problem either works or it doesn’t. And the more moral problems one solves the more one has progressed morally. That’s what I mean by morals are objective.
I don't think so. Morals are objective, and solutions to moral problems either work or they don't.
Besides, ethics as "the negotiation of conflicts of interest under conditions of shared purpose" sounds impressive (maybe) but it's vacuous.
When I was working on my last ebook, I found unsplash.com to be immensely helpful in finding beautiful, free-to-use, high-resolution images of all sorts. May even save you $$ if you can find a photo good enough for your cover so you don't need to hire a cover designer.
Simpler: ethics is the study or moral problems.
Better yet, it solves some of your own problems. If you’re both the creator and consumer, you can find and improve its flaws more easily, and your product will be so much better for it 💪
Custom implementation of 3 * 4 in JavaScript:
Array.from(Array(3).keys()).map(() => 4).reduce((acc, curr) => acc + curr);
Same in Berlin:
reduce(+ repeat(3 4))
Which do you prefer?
🙏
That's not to mention that wellbeing cannot be maximized because it can always get better.
I don't think slavery was abolished to maximize the well being of society. It was abolished because it is abhorrent. It was an instance of error correction—in this case a (grave) moral error—not an instance of optimization.
I'm afraid such a simulation is impossible because the main influencing factor—knowledge creation—is unpredictable in principle.
(It isn't really a problem anyhow!)
Well, it's like I said: through conjecture and criticism we can solve problems of all kind—be they scientific, moral, or otherwise. This interplay between conjecture and criticism is at the heart of Popperian epistemology, and it has enough reach to solve the is-ought problem.
Ah, I agree that science won't solve ethical problems, at least not most of the time. Moral problems are philosophical in nature, not scientific, yes.
Then presumably you think that abolishing slavery was no moral progress—because there is no such thing as moral progress?
I know it. I favor Popperian epistemology. The famous "you can't derive an ought from an is" is not a problem after all: as David Deutsch once said, we're not after deriving, we're after explaining! And explaining is done via conjecture, both for morals and otherwise.
I'm familiar with the work on "AI safety." Applying that stuff to narrow AI is morally ok, but applying it to AGI—who, by definition, are people (conscious, creative, etc)—is very sinister. Turns people into slaveholders.
Intelligence is the ability to solve problems. Including morals ones. These things are not orthogonal.
@HeuristicAndy
Boredom is your mind's way of telling you that something isn't for you. If someone thinks coding is dreadfully boring, they shouldn't do it.
Re GPT-3: Probably not.
When you put all of these things together: low bar to entry, good pay, beginner-friendliness, low cost, ability to make $$$ fast—how could you not recommend someone enduring hardship learn to code? It's their ticket out!
Seventh, the pay is good. Really good, even for beginners. I've never heard of a junior developer struggling to make ends meet. Your skills are too valuable for that to ever happen.
Related to that and sixth, because programming skills are incredibly valuable, you will rarely find yourself jobless, and if you do, it won't be for long. You can always find some job as a programmer to pay the bills.
Fifth, there are waaay more jobs than developers out there. Writing halfway decent JavaScript code will make $$$ within a few months of writing your first line. Getting that first paycheck thanks to a self-taught skill is one of the most empowering things you can experience.
Fourth, progress is rapid and transparent. Odds are whatever platform you're interested in working on gets improved at least every couple of months. That makes your life easier and increases your productivity. What other industry can say that?
Third, the industry is mostly open. You can find millions of lines of code for free, ask questions about it, tweak it, run it, tweak it again, contribute. The support is unlike any other and especially friendly to beginners who show they want to contribute and offer value.
That's part of what's so great about this industry: what matters is skills, not degrees. Talkers don't get far. Only qualified doers do.
I have been a software engineer for just short of ten years and never once has anyone asked me for any degrees. All my clients/employers ever wanted to know was wether I could solve a problem. Whenever I demonstrated I could, they hired me.
Second, you do not need a degree. In fact, I suggest you don't get one. You can learn everything you need online for free. And you can learn it in a matter of months. Compare that to a four-year degree and thousands of dollars of debt. It's a no-brainer.
First of all, coding is relatively easy. There, I said it. Granted, it's not the easiest thing in the world, but it's easier than many other career paths, esp. those involving manual labor, long hours at night, or work outside. That coding requiring genius is misleading folklore.
Detractors allege that telling someone who's in a bad spot to "learn to code" trivializes their hardship. But that's not at all the case: learning to code really is the best way to escape hardship. Why?
Actually, suggesting someone learn to code is great career advice.
A thread 🧵
What the let's-worry-about-AI folks don't know is: moral knowledge grows by the same logic as all other knowledge, and increased processing speeds make possible increased error correction of moral knowledge, too. So, we should give AGIs as much processing speed as possible!
Fairly smart, but not intelligent at all.
Yes, highly relevant for today's discussions with social-justice warriors.
Yes, automation frees people up to be more creative. By not having to execute tasks mindlessly, they can solve problems creatively.
"Machine speeds rather than human speeds"—the important constraining factor is going to be the performance characteristics of software, not hardware, as an ungainly and slow piece of software runs slowly even on a fast computer.
@SpaceX @NASA @AstroBehnken @Astro_Doug @Space_Station
Very nice. Congrats.
Saw this interesting question on Reddit: if Corona hadn't happened, how would the past ~4 months have played out differently for you?
I would have been able to go to the gym. And I probably would have eaten out more.
How would things have been different for you?
Exactamente. Tal vez le gusta también lo que dijo Popper de que todos somos iguales en nuestra ignorancia infinita.
@FranklinAmoo @JulienSLauret @TDataScience
Unless GPT-3 brings us closer in our understand of how the human mind works, it's not a step toward AGI. Nor can AGI be achieved through incremental steps—it requires something wholly new and qualitatively different.
By definition AGI and human intelligence are equivalent.
Agreed, because AGI cannot be achieved in steps—it requires something wholly new and qualitatively different. We won't be able to build it unless we understand how the human mind works.
Sí, el principio de optimismo probablemente es uno de los principios más importantes que conocemos. También su conjetura que problemas son solubles y inevitables.