When life gives you lemons, dig the sim cards out of them and get free internet.
376 karma · joined August 26, 2016
When life gives you lemons, dig the sim cards out of them and get free internet.
There's an old saying "If all you have is a hammer, all your problems start to look like nails." We may see an AI winter come about simply because we run out of things to pound with our hammer.
- Being results-oriented can be detrimental to doing anything useful.
As for how they are different from general dynamical systems, the GDSs I've seen have a small set of differential equations that describe the whole system; a cellular automaton has the same rule for each cell but the system comprises a large number of cells. In other words, locality and massive parallelism.
But this isn't my field so correct me if I'm wrong...
Or maybe contact O'Reilly and write an intro book "Semantic Web: Just the Good Parts" for their series.
As for semantic markup being confusing and usually wrong, I don't know where you get that.
I think he's applying closed source's rules. A closed source project is a failure if no one uses (ie, buys) it, since it represents a development cost that wasn't compensated. Not so with open source. I don't suffer if no one uses my project, and the existence of my repositories doesn't detract from anyone else's work.
It's difficult to evaluate CYC given Cycorp's secrecy. Domingos called it something like a colossal failure, which is hard to believe, but it's hard to argue given how little info gets out of Cycorp. So does Cycorp have any interest in changing this? Do they care about improving their reputation in the AI community?
How would you compare Cyc (both the knowledge base and the inference engine) with current efforts in the semantic web? Is OWL/RDF just too simple to encode the kind of logic you think common sense needs?
I am pleased to see paperswithcode.com as a collection point of research. But the fact that it casts all of AI as one big set of contests on datasets where "progress" is equated with small increments of classification accuracy is, well, a disappointing view of the field.
> Sure; all the parts haven't been completely connected to make self driving cars, general game playing AIs or conversation bots.
This implies that connecting these systems is just a detail to be worked out. But the systems have nothing in common ontologically---they share no inputs, outputs, or intermediate concepts. The fact that they both use neurons doesn't mean much. You may as well say we can connect snails and dogs.
> the biggest rise in prescriptions during this time period was for back pain and other types of chronic pain.
Now I'm doubly surprised. Some benzos have muscle relaxant properties but I've never heard of them prescribed for pain.
(I've never experienced benzo withdrawal but I've heard it claimed to be as bad as opiate withdrawal.)
As for Valium, it seems odd this article even makes a big deal about it. Valium (and benzodiazepines in general) aren't prescribed as much these days because of their addictive potential. Their spike of popularity came and went in a few decades, roughly 1960-1980. Benzos are still popular in illicit use, but a key point of Soma was that it wasn't illicit---it was sanctioned by the state.
I don't recall ever reading in a technical paper, or in an interview, a leader in the field of ANNs claim they were thinking. If you have, I'd like to see a reference. Most are fairly honest about the differences between artificial neurons and real ones, and between human cognition and what ANNs are doing with data.
1. The difference between collecting bus tickets, working on mathematics, and figuring the crystallization patterns of snowflakes is virtually nothing. To the people obsessed with it, it is all-important; to outsiders, all three look indistinguishably pointless. Only in retrospect can anyone say whether the activity led to something society considers important. The truly obsessed don't care.
2. Genius is overrated. We like stories of great individuals changing the world with their genius because we like stories and we dislike chance. But for nearly every invention and idea, history shows that multiple people were working on it at the same time. Civilization's progress is more a sequence of ideas whose time had come, but we prefer stories of great unique geniuses.
To quote the article, "Model-based reinforcement learning aims to address this issue by first learning a model of the environment’s dynamics, and then planning with respect to the learned model."
So they've sped up learning from examples by learning a model first, but it's still learning from examples.
Your examples (deep learning applied to perception) are what he argues AI is generally good for.
This may be his biggest impediment. ML has gotten very far with looking at problems as linear algebraic systems, where optimizing a loss function mathematically yields a good solution to a precisely defined (and well circumscribed) classification or regression problem. These techniques are very seductive and very powerful, but the problems they solve have almost nothing in common with AGI.
Put another way, Machine Learning as a field diverged from human learning (and cognitive science) decades ago, and the two are virtually unrecognizable to each other now. Human learning is the best example of AGI we have, and using ML tech as a way to get there may be a seductive dead end.
Eh? It's well known in Info Retrieval; it goes back at least to Salton's IR group at Cornell in the 1970's.
It is slightly better now than it used to be, i.e. you now have to read two sentences to realize it's incoherent gibberish, rather than just one. But thinking you're getting closer to understanding this way is like the metaphor of climbing a tree to reach the moon: you can keep reporting progress until you run out of tree.
As for it being entertaining, sure, the first attempts were entertaining and instructive. Now, years later, reading more gibberish from a neural net on HN, not so much.
Instead of asking about values, ask about behavior. No one is going to say no to "Do you value code quality?" Aking what they've done (and sacrificed) to support code quality will give you a better idea of what their relative values are.
Asking about their tech ladder will get you an HR answer. Asking "Who was the last person promoted to X? When?" will give you a better answer. And if they can't remember the last time, that's an answer in itself.
Related to time anxiety, Oliver Burkeman pointed out the paradox of "making good use of time". Taken too far, it turns time into an instrumental goal. You never really enjoy it; you're always just investing time in a future that never comes. And you suffer time anxiety.