← Feed

Posts

• Post #4450557

  I do hate how the recent oppression of trans people is often presented as being about "biological sex." It's absurd to claim that all of human sex and gender can be reduced to a binary. I mean, look at it! It's weird, complicated, and incredibly diverse! No, it's just an attempt to sound justified by using sciencey, authoritative words. What they really mean is: you were assigned a role, and you are expected it to play it, setting all biology, feelings, autonomy, and dig...

• Post #4377185

  John Oliver just did a great piece on police surveillance. You know, I love machine learning as a technology. It's a really fascinating, useful tool. But I hate how so many people treat it as magically reliable. That's just foolish. Modern computer vision often achieves 90+% accuracy in ideal conditions, and... that's pretty mediocre, actually! It's good as a first pass filter, or a "good enough" solution to something low stakes, but it's absolutely insane to me th...

• Post #2959256

  I've got a quick question for the friendly microbiologists of Mastodon. It's my understanding that it's common for some environmental change to wipe out 90+% of a bacterial colony, only to have the survivors quickly multiply back to their original numbers, this time with a mutation that grants resistance to whatever the threat was. Is there a name for this phenomenon (not "gene fixation," but the die back and recovery)? Are there canonical papers about it? EDIT: I think...

• Post #2959255

  Man, marsupials are so weird. I just found out they don't have a corpus callosum connecting the two halves of their brain, and I was like: of course. Why not. Very marsupial. Two opossums, one body. Never change, marsupials.

• Post #2959254

  It's obnoxious how every website wants you to install their app or whatever. Like, the whole point of the web is that you can access any information and software without installing it on your computer. The modern browser is a sandbox and a virtual machine, a portal that grants you access to any site without granting them access to you (in theory). But everyone wants access to your machine. They want to stay resident, to monitor and monetize you, even if it's a security vulnerability an...

• Post #2959253

  I hate how LLMs are often used, but as a model of cognition I think they're fascinating. Take this paper, for example. Here they look at reasoning over long sequences of data that can't fit into working memory. The latest LLMs can solve problems like this, using hybrid memory models that consolidate "short-term" memories in an LRU cache into "long-term" memories in the form of persistent weights (which is good to know about generally). However, this paper shows tha...

• Post #2558220

  My research is about how life evolved the mechanisms for evolution as we see today, and how the particular form they take shapes the direction and speed of evolution. Interestingly, that's led me to the origin of life literature! It turns out, biologists are mostly concerned with evolution as it is today, which makes it hard to talk about how the process of evolution itself evolved. So, I'm exploring papers like this one that talk explicitly about the emergence of genetics and how that...

• Post #2535667

  Neural networks have been one of the most transformative technologies of the past decade. They were originally inspired by human brains, and their promise is to help us understand human intelligence (or replace it, depending on who you ask). With all the excitement and drama around AI these days, you might be forgiven for thinking we're nearly there! In reality, though, we've only understood a tiny fraction of what the brain is doing. I'd like to convince you that what's happenin...

• Post #2394650

  A lab mate recently asked me why I think the tech industry might regret their rapid and large scale adoption of LLM into their development practices. I said there were at least three core problems: tech companies outpacing themselves, reduced quality, and cultural breakdown. (1/4) #ai #llm #programming

• Post #2142635

  Has anyone ever seen code reviews working well in an academic setting? One of my lab mates just paid a very high price for a small bug, and we're talking about using code reviews to prevent this. Coming from industry myself, I like this idea and volunteered to help figure this out, but I can also imagine lots of reasons this might not work in a lab full of grad students and postdocs. In particular, I'm worried about training people to actually give good code reviews, sharing the workl...

• Post #1543749

  The most important "productivity hack" I have learned is to recognize when my brain is out of juice for the day. It has a very distinct feeling to it. Once that happens, no work of quality or substance will get done, no matter how long I bang my head against it. So, I might as well go home and rest. I can't express just how much my well-being improved after I fully embraced this. And I get more done, of higher quality.

• Post #1490152

  I enjoy puzzles, but there are many kinds, of course. I think I'm drawn to science and engineering because I like "why is this not working?" puzzles. Those are some of the most frustrating puzzles I know! Also the most rewarding. I wonder how much of a factor this sort of thing is in what sort of jobs and lifestyles people choose for themselves?

• Post #1490151

  I just started reading From Deep Learning to Rational Machines: What the History of Philosophy Can Teach Us about the Future of Artificial Intelligence by Cameron Buckner. I can already tell this is going to be an interesting read for me. I'm quite frustrated by how much DL research seems to think getting high scores on benchmarks means they have "solved intelligence," without really stopping to think about what they've actually done and how it relates to what we call "i...

• Post #1490146

  I like watching Last Week Tonight, although it is often upsetting. The most recent episode on Prediction Markets hit me a lot harder than I expected. I dunno. Clearly there are worse things going on in the world. But the sort of hyper-capitalist nihilism these apps represent is profoundly depressing to me. It's even worse to see this being promoted by the Trump administration and adopted by young people. "The world is burning, might as well make a quick buck off of others' misfort...

• Post #1423704

  These days, so much of our lives and work and relationships are mediated by screens. I think this is a big part of why folks think AI is approaching human capacity. It's important to remember: we choose to interact through this narrow, impoverished interface. It is not normal or healthy for human beings, just convenient for our mechanized society. It limits us, captures only a small sliver of our humanity, and makes building relationships hard. Deep learning can do amazing things with tex...

• Post #1405059

  I thought this was a particularly good analysis of the problem of using LLMs for science. It explores the purpose of science, the perverse incentives that drive people to use LLMs, and the impact this has on skill building and training future scientists. My lab group has been struggling with this topic lately, without much consensus. This blog post captures a lot of our thinking, and very clearly made some good points that we appreciated. It mostly just describes the mess we're in without...

• Post #1049304

  I just had to buy on over-the-counter home medical test kit, and the only option they had was a digital one. That means they increased the cost of the kit to put in disposable electronics (destined to become e-waste) so they can force me to install an app (who knows what tracking they're doing?) and endure ads for telehealth and prescription drugs just to get the results. WTF?! This is disgusting, and should 100% be illegal.

• Post #1049302

  Welch Labs puts out some killer AI explainer videos! I really enjoyed this one about using LLMs for robotics. It explores the VLA architecture, which cleverly integrates latent spaces between natural language, computer vision, and motion planning tasks. This is an elegant algorithm, and I feel like the video makes it very intuitive for me why it works so well. I'm also just glad to see more well explained examples of transformer architectures being used in interesting ways and going beyond...

• Post #768242

  "In this work, we demonstrate that LLMs not only alter the voice and tone of human writing, but also consistently alter the intended meaning." "heavy LLM users reported that the writing was less creative and not in their voice." "Even when LLMs are prompted with expert feedback and asked to only make grammar edits, they still change the text in a way that significantly alters its semantic meaning." "the LLM is not merely correcting grammar, but is actively s...

• Post #762179

  My friend seems genuinely baffled that I am an AI researcher who refuses to use AI! Not only that, but I argue against it from theory, not experience. Why don't I just give it a try for a while, and see what it's really about before I judge it? I guess I see where he's coming from. Part of the problem is the word "AI." LLMs are not my research focus, so it's less of a contradiction than it sounds. But I admit, being a non-user makes my arguments against LLMs less credi...

• Post #459761

  Ugh, LinkedIn is the worst! I try to disable all the notifications that aren't directly relevant to me, but they keep inventing new kinds of notifications that I have to opt out of! So annoying. Trying to do that this morning, I see they have "simplified and regrouped" their notification settings, which is hilarious, because they're showing me a list of 14 top-level notification categories, each with its own tree of sub-categories beneath it. There must be a few dozen differe...

• Post #433544

  I'm so confused. I just found a small body of literature applying LLMs to reinforcement learning type tasks, exploring the use of LLMs for "autonomous decision making." I guess people are building more LLM agent systems, and we ought to understand them and what makes them better / worse at what they do. But I still feel like LLMs are fundamentally not suited to decision making tasks. They don't weigh options and decide. At best, you could say they interpolate what a reasonabl...

• Post #366777

  Good AI research should tell us something about life, or it should help people. I hate seeing research about automating what people do. It's not a good goal for science or society! I was recently reminded of this by a paper applying LLMs to math. This domain has many good questions: what do we mean when we say a person "solves math problems"? What are they actually doing? How is this like or not like what an LLM does? How might mathematicians benefit from this? Instead, we get pa...

• Post #366776

  These days it sure feels like the tech industry is aggressively pushing all computing, hardware and software, onto a service / subscription model. I find this particularly offensive because I know many of the executives doing this used to be engineers, who honed their craft on affordable personal computers. They're denying the next generation the opportunity that made them successful. They're not just gatekeeping access to computers, but also to making better alternatives. They're...

• Post #366774

  I really enjoyed this Artificial Life paper. Beer's lab focuses on finding ways to describe the remarkable things life does in formal terms, and build tools to visualize and analyze them. That's a really cool mission, I think! In this case, the focus is on what makes an individual more or less "viable" or "adaptable" than another one. They did this in a really simple ALife setting based on chemical diffusion. They identified self-sustaining patterns, then tried poki...

• Post #366773

  I got invited to update my last paper with some extra material. So, I went back and started probing deeper, and it has been a journey! Wait, is this really doing what I thought? If it's not, does that undermine everything I was saying? Or can both stories be true? Is this even interesting any more? Is there a real effect, or is it just an illusion? It is behaving systematically, though, with measurable effects. So I can identify the principles... And I can use that to propose a better,...

• Post #366772

  Reinforcement Learning is AI for playing games, planning actions, and doing things in the real world. It's extremely powerful, but also a fun and interesting topic to learn. I'm teaching myself now. I was always intimidated by all the esoteric jargon and math in RL. The classic Sutton and Barto text clears that up nicely, though. It's one of the best text books I have ever studied, and it reveals RL is much less complex than it sounds. I think the excessive mathiness comes from the...

• Post #366771

  Why should I learn to code, when Claude could do it for me? I realized recently, this is just like asking: Why should I learn Spanish, when I could use Google Translate? There are many reasons. A language is vastly richer than any translation. Becoming fluent lets you express yourself more eloquently, and in new ways. It gives you access to new perspectives, and new tools for thought and creativity. It helps you connect with a community of people like you, and all of the works they've...