Press "Enter" to skip to content

Posts published in “Intelligence”

Artificial intelligence: making computers do things we’d call intelligent if done by people.

[twenty twenty-four day forty-three]: neurodivergence by the numbers

centaur 0

So! While working on The Neurodiversiverse I've been reading up a lot on neurodiversity. According to Devon Price's Unmasking Autism, autism is massively undiagnosed, and for good---well, understandable---reasons. From parents concerned about their uncommunicative children or fans of cold geniuses on Sherlock and the Big Bang Theory, our culture focuses a lot on certain stereotypes of autism---while ignoring a much larger group of people who suffer from the same underlying conditions in their brains, but who are able to "mask" their behavior to appear much more "high functioning" or even "neurotypical".

As you might imagine, spending your whole day trying to react in ways that are fundamentally unnatural to you---and trying to hide the ways that you react that are natural to you---can stress people the fuck out. But many people never get a diagnosis---either because they're from a disadvantaged group, or because they don't want to risk the stigma and potential negative consequences of a diagnosis, or because they mask too well and no-on notices how they are suffering. But if you don't understand your condition, you may employ coping strategies which may actually do more long-term harm than good.

Well, now there are a lot of online tests and self-help books and even sympathetic therapists who can help people understand themselves better. While I've always known I was a bit strange---mostly solitary, typically withdrawn at family gatherings when I was a child, or explicitly labeled as having a weird brain---I've never pursued a diagnosis of any kind---in the past, because I didn't feel I had any trouble coping to the point that I needed help, and in the present, because having a disability label attached to you can have negative social and legal consequences that I have no interest in dealing with.

BUT! The personal stories of Unmasking Autism resonated a lot with me, and I now have friends who have gone through formal adult diagnoses of autism and ADHD, as well as an undiagnosed autistic friend who clearly is autistic and has to manage her life the way a masking autistic person does, but who did not pursue a diagnosis for precisely the same reasons that many other masking autistics do not pursue it: unless your condition is very severe, it isn't clear that a formal diagnosis can actually get you help, and it can often get you a lot of hurt. But UNDERSTANDING it, that, that we can now do.

So! And I note I again use "So!" at the start of a paragraph. Is that a verbal tic? Who cares? SO ANYWAY ...

Diagnoses of autism, and other neurodivergences! The neurodivergence I identify most with is Social Anxiety Disorder---in fact, this is the neurodivergence I chose for the protagonist of "Shadows of Titanium Rain", my own submission to The Neuroversiverse. But other people have suggested I have characteristics of OCD, or ADHD, or Autism, and I even went into therapy for stress and anxiety during the pandemic. So I decided to take five online tests: Social Anxiety Disorder, Autism, Anxiety, ADHD, and OCD.

The results are at the top of the blog---and I already gave away the game through the order I listed them. Normalizing all the scores from zero to a hundred, most of the tests put the boundary of "you've got the thing" at somewhere around 60-70% of the possible points you could score - let's call it at 2/3, or 66%, shall we? OCD scored the lowest - roughly 53%, which the test judged as "you've got OCD tendencies, but not OCD." ADHD was a little higher, 60%, and general Anxiety still higher, 63%. But none of these were over the "you've got it" thresholds for these tests---they just indicated a general tendency in that direction.

Things start to change with Autism: my test results for "Adult Autism" (*cough* MISNOMER) were 70%, well within the boundary of "you've very probably got it". Some of my friends are quite surprised to hear this, as they didn't see this in me at all; I guess my condition is "mild" and/or I mask very well.

But Social Anxiety Disorder? 86%, off the charts. And this wasn't a surprise: not only do I have a huge raft of coping mechanisms to help me deal with social situations, I also have some of the more subtle symptoms of Social Anxiety Disorder that you might not expect would be symptoms. For example, in certain socially awkward situations, I can partially stumble while walking. Most people, even those close to me, never notice that my foot briefly drags when we're walking and something socially awkward occurs - yet balance and coordination issues are a symptom of social anxiety.

Again, I've not pursued a formal diagnosis, and I don't plan to. But understanding these things about myself helps me understand why I've built a mass of coping mechanisms and masking strategies in my life---and can help me start to construct a healthier way to cope with the world within which I live.

If you feel alienated by your world, perhaps that's something you could try too.

-the Centaur

Robots!

centaur 0

Still at the Conference on Robot Learning. LOTS of robot dogs were about, lots of diffusion model and transformer work, and lots of language model planning. More later, gotta crash.

-the Centaur

Announcing Logical Robotics

centaur 0

So, I'm proud to announce my next venture: Logical Robotics, a robot intelligence firm focused on making learning robots work better for people. My research agenda is to combine the latest advances of deep learning with the rich history of classical artificial intelligence, using human-robot interaction research and my years of experience working on products and benchmarking to help robots make a positive impact.

Recent advances in large language model planning, combined with deep learning of robotic skills, have enabled almost magical developments in explainable artificial intelligence, where it is now possible to ask robots to do things in plain language and for the robots to write their own programs to accomplish those goals, building on deep learned skills but reporting results back in plain language. But applying these technologies to real problems will require a deep understanding of both robot performance benchmarks to refine those skills and human psychological studies to evaluate how these systems benefit human users, particularly in the areas of social robotics where robots work in crowds of people.

Logical Robotics will begin accepting new clients in May, after my obligations to my previous employer have come to a close (and I have taken a break after 17 years of work at the Search Engine That Starts With a G). In the meantime, I am available to answer general questions about what we'll be doing; if you're interested, please feel free to drop me a line at via centaur at logicalrobotics.com or take a look at our website.

-the Centaur

phewww ….

centaur 0

... finally, a chance to catch a break.

It's been a difficult few weeks due to "the Kerfluffle" which I hope to blog about shortly (those on my LinkedIn have seen it already) but equally as much from a Stanford extension class I was taking on Deep Reinforcement Learning (XCS234 - speaking as an expert in this area seeking to keep my skills sharp, I can highly recommend it: I definitely learned some things, and according to the graphs, so did my programs).

Finally, that's over, and I have a moment to breathe.

And maybe start blogging again.

-the Centaur

Pictured: A mocha from Red Rock Cafe, excellent as always, and a learning curve from one of my programs from class (details suppressed since we're not supposed to share the assignments).

Ugh, WordPress updates edition …

centaur 0

... the block editor of Wordpress seems to be making my old non-block-editor posts turn into solid walls of text. See the post "Pascal's Wager and Purchasing Parsley":

Yeah, it's not supposed to be looking like that. Gotta track those down and fix them.

In other news, my Half-Cheetah policy is successfully training to "expected" levels of performance. Yay! I guess that means my code for the assignment is ... sorta correct? Time to clean it up and submit it.

-the Centaur

Once again, I’m running deep learning on a Macbook …

centaur 0

... and the trick to getting it working was, as usual, "working just a little bit harder than you want to". Shortly after my last post, I got REINFORCE, a classic reinforcement learning algorithm, successfully training on my local machine, with apparent learning for all three environments in the assignment (though whether my solution is able to reach the expected final level of performance or not is still an open question).

-the Centaur

The Embodied AI Workshop is Tomorrow, Sunday, June 20th!

centaur 0
embodied AI workshop

What happens when deep learning hits the real world? Find out at the Embodied AI Workshop this Sunday, June 20th! We’ll have 8 speakers, 3 live Q&A sessions with questions on Slack, and 10 embodied AI challenges. Our speakers will include:

  • Motivation for Embodied AI Research
    • Hyowon Gweon, Stanford
  • Embodied Navigation
    • Peter Anderson, Google
    • Aleksandra Faust, Google
  • Robotics
    • Anca Dragan, UC Berkeley
    • Chelsea Finn, Stanford / Google
    • Akshara Rai, Facebook AI Research
  • Sim-2-Real Transfer
    • Sanja Fidler, University of Toronto, NVIDIA
      Konstantinos Bousmalis, Google

You can find us if you’re signed up to #cvpr2021, through our webpage embodied-ai.org or at the livestream on YouTube.

Come check it out!

-the Centaur

He thinks he invented Java because he was in the room when someone made coffee

taidoka 0

... came up as my wife and I were discussing the "creative hangers-on form" of Stigler's Law. The original Stigler's Law, discovered by Roger Merton and popularized by Stephen Stigler, is the idea that in science, no discovery is named after its original discoverer.

In creative circles, it comes up when someone who had little or nothing to do with a creative process takes credit for it. A few of my wife's friends were like this, dropping by to visit her while she was in the middle of a creative project, describing out loud what she was doing, then claiming, "I told her to do that."

In the words of Finn from The Rise of Skywalker: "You did not!"

In computing circles, the old joke referred to the Java programming language. I've heard several variants, but the distilled version is "He thinks he invented Java because he was in the room when someone made coffee."  Apparently this is a good description of how Java itself was named, down to at least one person  claiming they came up with the name Java and others disputing that, even suggesting that they opposed it, claiming instead that someone else in the room was responsible - while that person in turn rejected the idea, noting only that there was some coffee in the room from Peet's.

Regardless, I dispute Howard Aiken's saying "Don't worry about people stealing your ideas. If your ideas are any good, you'll have to ram them down people's throats." Nah. Once you've forced an idea down someone's throat, they won't just swallow it, they'll claim it was in their stomach all along.

-the Centaur

The Embodied AI Workshop at CVPR 2021

centaur 0
embodied AI workshop

Hail, fellow adventurers: to prove I do something more than just draw and write, I'd like to send out a reminder of the Second Embodied AI Workshop at the CVPR 2021 computer vision conference. In the last ten years, artificial intelligence has made great advances in recognizing objects, understanding the basics of speech and language, and recommending things to people. But interacting with the real world presents harder problems: noisy sensors, unreliable actuators, incomplete models of our robots, building good simulators, learning over sequences of decisions, transferring what we've learned in simulation to real robots, or learning on the robots themselves.

interactive vs social navigation

The Embodied AI Workshop brings together many researchers and organizations interested in these problems, and also hosts nine challenges which test point, object, interactive and social navigation, as well as object manipulation, vision, language, auditory perception, mapping, and more. These challenges enable researchers to test their approaches on standardized benchmarks, so the community can more easily compare what we're doing. I'm most involved as an advisor to the Stanford / Google iGibson Interactive / Social Navigation Challenge, which forces robots to maneuver around people and clutter to solve navigation problems. You can read more about the iGibson Challenge at their website or on the Google AI Blog.

the iGibson social navigation environment

Most importantly, the Embodied AI Workshop has a call for papers, with a deadline of TODAY.

Call for Papers

We invite high-quality 2-page extended abstracts in relevant areas, such as:

  •  Simulation Environments
  •  Visual Navigation
  •  Rearrangement
  •  Embodied Question Answering
  •  Simulation-to-Real Transfer
  •  Embodied Vision & Language

Accepted papers will be presented as posters. These papers will be made publicly available in a non-archival format, allowing future submission to archival journals or conferences.

The submission deadline is May 14th (Anywhere on Earth). Papers should be no longer than 2 pages (excluding references) and styled in the CVPR format. Paper submissions are now open.

I assume anyone submitting to this already has their paper well underway, but this is your reminder to git'r done.

-the Centaur

A Bayesian Account of Miracles

centaur 0
bayes headshot

Christianity is a tall ask for many skeptically-minded people, especially if you come from the South, where a lot of folks express Christianity in terms of having a close personal relationship with a person claimed to be invisible, intangible and yet omnipresent, despite having been dead for 2000 years.

On the other hand, I grew up with a fair number of Christians who seem to have no skeptical bones at all, even at the slightest and most explainable of miracles, like my relative who went on a pilgrimage to the Virgin Mary apparitions at Conyers and came back "with their silver rosary having turned to gold."

Or, perhaps - not to be a Doubting Thomas - it was always of a yellowish hue.

Being a Christian isn't just a belief, it's a commitment. Being a Christian is hard, and we're not supposed to throw up stumbling blocks for other believers. So, when I encounter stories like these, which don't sound credible to me and which I don't need to support my faith, I often find myself biting my tongue.

But despite these stories not sounding credible, I do nevertheless admit that they're technically possible. In the words of one comedian, "The Virgin Mary has got the budget for it," and in a world where every observed particle event contains irreducible randomness, God has left Himself the room He needs.

But there's a long tradition in skeptical thought to discount rare events like alleged miracles, rooted in  Enlightenment philosopher David Hume's essay "Of Miracles". I almost wrote "scientific thought", but this idea is not at all scientific - it's actually an injection of one of philosophy's worst sins into science.

Philosophy! Who needs it? Well, as Ayn Rand once said: everyone. Philosophy asks the basic questions What is there? (ontology), How do we know it? (epistemology), and What should we do? (ethics). The best philosophy illuminates possibilities for thought and persuasively argues for action.

But philosophy, carving its way through the space of possible ideas, must necessarily operate through arguments, principally verbal arguments which can never conclusively convince. To get traction, we must move beyond argument to repeatable reasoning - mathematics - backed up by real-world evidence.

And that's precisely what was happening right as Hume was working on his essay "Of Miracles" in the 1740's: the laws of probability and chance were being worked out by Hume's contemporaries, some of whom he corresponded with, but he couldn't wait - or couldn't be bothered to learn - their real findings.

I'm not trying to be rude to Hume here, but making a specific point: Hume wrote about evidence, and people claim his arguments are based in rationality - but Hume's arguments are only qualitative, and the quantitative mathematics of probability being developed don't support his idea.

But they can reproduce his idea, and the ideas of the credible believer, in a much sounder framework.

In all fairness, it's best not to be too harsh with Hume, who wrote "Of Miracles" almost twenty years before Reverend Thomas Bayes' "An Essay toward solving a Problem in the Doctrine of Chances," the work which gave us Bayes' Theorem, which became the foundation of modern probability theory.

If the ground is wet, how likely is it that it rained? Intuitively, this depends on how likely it is that the rain would wet the ground, and how likely it is to rain in the first place, discounted by the chance the ground would be wet on its own, say from a sprinkler system.

In Greenville, South Carolina, it rains a lot, wetting the ground, which stays wet because it's humid, and sprinklers don't run all the time, so a wet lawn is a good sign of rain. Ask that question in Death Valley, with rare rain, dry air - and you're watering a lawn? Seriously? - and that calculus changes considerably.

Bayes' Theorem formalizes this intuition. It tells us the probability of an event given the evidence is determined by the likelihood of the evidence given the event, times the probability of the event, divided by the probability of the evidence happening all by its lonesome.

Since Bayes's time, probabilistic reasoning has been considerably refined. In the book Probability Theory: The Logic of Science, E. T. Jaynes, a twentieth-century physicist, shows probabilistic reasoning can explain cognitive "errors," political controversies, skeptical disbelief and credulous believers.

Jaynes's key idea is that for things like commonsense reasoning, political beliefs, and even interpreting miracles, we aren't combining evidence we've collected ourselves in a neat Bayesian framework: we're combining claims provided to us by others - and must now rate the trustworthiness of the claimer.

In our rosary case, the claimer drove down to Georgia to hear a woman speak at a farmhouse. I don't mean to throw up a stumbling block to something that's building up someone else's faith, but when the Bible speaks of a sign not being given to this generation, I feel like its speaking to us today.

But, whether you see the witness as credible or not, Jaynes points out we also weigh alternative explanations. This doesn't affect judging whether a wet lawn means we should bring an umbrella, but when judging a silver rosary turning to gold, there are so many alternatives: lies, delusions, mistakes.

Jaynes shows, with simple math, that when we're judging a claim of a rare event with many alternative explanations, our trust in the claimer that dominates the change in our probabilistic beliefs. If we trust the claimer, we're likely to believe the claim; if we distrust the claimer, we're likely to mistrust the claim.

What's worse, there's a feedback loop between the trust and belief: if we trust someone, and they claim something we come to believe is likely, our trust in them is reinforced; if we distrust someone, and they claim something we come to believe is not likely, our distrust of them is reinforced too.

It shouldn't take a scientist or a mathematician to realize that this pattern is a pathology. Regardless of what we choose to believe, the actual true state of the world is a matter of natural fact. It did or did not rain, regardless of whether the ground is wet; the rosary did or did not change, whether it looks gold.

Ideally, whether you believe in the claimer - your opinions about people - shouldn't affect what you believe about reality - the facts about the world. But of course, it does. This is the real problem with rare events, much less miracles: they're resistant to experiment, which is our normal way out of this dilemma.

Many skeptics argue we should completely exclude the possibility of the supernatural. That's not science, it's just atheism in a trench coat trying to sell you a bad idea. What is scientific, in the words of Newton, is excluding from our scientific hypotheses any causes not necessary or sufficient to explain phenomena.

A one-time event, such as my alleged phone call to my insurance agent today to talk about a policy for my new car, is strictly speaking not a subject for scientific explanation. To analyze the event, it must be in a class of phenomena open to experiments, such as cell phone calls made by me, or some such.

Otherwise, it's just a data point. An anecdote, an outlier. If you disbelieve me - if you check my cell phone records and argue it didn't happen - scientifically, that means nothing. Maybe I used someone else's phone because mine was out of charge. Maybe I misremembered a report of a very real event.

Your beliefs don't matter. I'll still get my insurance card in a couple of weeks.

So-called "supernatural" events, such as the alleged rosary transmutation, fall into this category. You can't experiment on them to resolve your personal bias, so you have to fall back on your trust for the claimer. But that trust is, in a sense, a personal judgment, not a scientific one.

Don't get me wrong: it's perfectly legitimate to exclude "supernatural" events from your scientific theories - I do, for example. We have to: following Newton, for science to work, we must first provide as few causes as possible, with as many far-reaching effects as possible, until experiment says otherwise.

But excluding rare events from our scientific view of the world forecloses the ability of observation to revise our theories. And excluding supernatural events from our broader view of the world is not a requirement of science, but a personal choice - a deliberate choice not to believe.

That may be right. That may be wrong. What happens, happens, and doesn't happen any other way. Whether that includes the possibility of rare events is a matter of natural fact, not personal choice; whether that includes the possibility of miracles is something you have to take on faith.

-the Centaur

Pictured: Allegedly, Thomas Bayes, though many have little faith in the claimants who say this is him.