The Stack Overflow Podcast

Letting algorithms guide our path to the next great invention

Episode Summary

The home team discusses a new alternative to smartphones, how AI will impact scientific research and journalism, another dispiriting round of layoffs in tech, and building a computer with Legos.

Episode Notes

Rabbit R1 is an AI-powered assistant you can keep in your pocket (but it’s not a phone).

How will AI impact scientific research? A new collaboration between Microsoft and the Pacific Northwest National Laboratory is focused on energy storage solutions.

A US Senate hearing questions whether tech companies should be allowed to train their AI models on content produced by journalists without paying licensing fees.

Learn how to build a mechanical computer from Legos.

Episode Transcription

[intro music plays]

Ben Popper Hello, everybody. Welcome back to the Stack Overflow Podcast, a place to talk all things software and technology. I am your host, Ben Popper, Director of Content here at Stack Overflow, joined by my wonderful teammates from the content crew, Ryan Donovan and Eira May. Hi, y'all. 

Ryan Donovan Hey. How’re you doing? 

Eira May Hey, folks. How's it going? 

BP It's going pretty well. So I want to kick us off with some fun stuff. I used to go every year to CES, the Consumer Electronics Show, in Las Vegas when I worked as a tech reporter at The Verge. It took place this week and a lot of fun stuff came out of it, but one that caught my eye was called The Rabbit. It's from the folks at Teenage Engineering, who, if you're listening to this podcast, you may know them. They're kind of an artisanal gadget shop, I want to say. They make things that they know aren't going to sell a lot, they're not mass produced. They're sort of labors of love, whether it's a keyboard or a gaming device or whatever it may be. And so this one is kind of like that company Humane. It's an alternative to having a phone. You have a screen almost like a Tamagotchi. It's got a cute little rabbit on it, and you can talk to it and it will answer back with the Gen AI smarts about any topic you want to discuss, and even, they said, handle a few tasks like book an Uber, look up a recipe, et cetera, et cetera.

RD What's the advantage of this over, say, a smartphone, or that pin we talked about a few weeks ago?

BP I think that there is no advantage in a base utility sense. I think maybe you want to give this to your kids so they can always ask interesting questions, but not have a phone. Maybe you want to spend some time away from a phone. I increasingly feel that by the end of the day I have eye strain and I'm hoping to invest in some stronger blue light glasses. And then, like I said, it's also kind of a gadget for the gadget lover. What's the point of having a Tamagotchi? They're hoping that you'll build some maybe sentimental attachment to this little rabbit in your pocket that answers all your questions, I guess. 

RD It's a lot cheaper than a phone. 

EM I like that people build a sentimental attachment to their phones over time. 

RD Is this the pet rock for the AI era?

BP Yeah, I think given Humane and given this, we're going to see a lot of little Gen AI assistants in x container. You know that's going to be in a Ring. It's going to be in a pair of glasses. You're going to have a little Her voice in your ear in a lot of different ways. 

RD I said earlier, this is the new Doom. It’s going to be running on a pregnancy test. 

BP Yeah, exactly. Can you run this on x? I've really gotten into a habit recently in the morning before I've had my coffee and my kids are asking me the kind of inane questions that they do, and I could try to make up the answer to this, but if you just ask ChatGPT, this conversation can go on as long as you like and I can have my coffee. So we're thinking about going to Iceland. They're like, “How does the aurora borealis work?” And I'm like, “I could make this up, but ask the oracle,” and then it's like, “Ooh, the sun, free radicals, magnetic fields,” and it feels like they're learning something. 

RD That's good. It's getting them to learn a little something. I guess it's better than unleashing them on the raw internet. 

EM Yeah, it’s nice to know where the floor is. 

BP I can say that I definitively agree. 

RD The aurora borealis is just chemtrails. 

BP Ooh, look into it. 

EM Hot take from Ryan. 

RD That's right. Got to get those conspiracies going.

BP All right, I'm going to move us on to a new topic that I really enjoy. One of the interesting things that comes up as we discuss AI reaching these new capabilities is that it can have a conversation with you, but can it generate anything meaningfully new or novel? And when it comes to LLMs, I think the debate is out about that. But Microsoft shared a long video earlier this week where they talked about a scientific collaboration that came up with a new sort of concept for a battery that would use far less lithium. It’d be a much more efficient battery, and in this way would be way better for both industry and for the environment. And the design of the approach to this was generated by an AI that was taught, “Hey, here's how you conceive of a novel material or a novel chemical interaction. Can you go out and do that and see what you can cook up on this?” And then it iterates and just goes on and on and on and eventually it stumbles on something pretty cool. So I don't want to get out of control here, Ryan, but the idea that we can have beneficial scientific discoveries, drug discovery for new medicines, material or design discovery to reduce pollution, that seems like a real upside to AI. What do you think? 

RD I think folks have been using AI for these sorts of things. There was the protein folding one where I think you would just run it on your computer and it would just simulate protein folding to see if anything was interesting. And I don't know how much they're using AI, but folks at IBM Quantum, when we spoke to them, they're doing a lot of material simulation in terms of chemistry. So I think AI figuring out how to implement chemical properties better, that's great. 

BP Yes, this news was, as you said, attached to the Microsoft Quantum team. There isn't really a lot of utility for these quantum supercomputers yet out in industry, but they claim it can winnow down 32 million potential inorganic compounds to 18 promising candidates in 80 hours, and that work would take human beings several years. So that's kind of a fun way to use it and help us move faster in identifying promising scientific paths of exploration.

RD Right, and then the scientists can take those 18 or whatever and do the actual science on them and prove it out in the real world.

EM Doing science. 

BP Do the science, as we say. All right, Eira. Why don't you bring up the next one, the Ars Technica link you shared.

EM So there's senate hearings going on this week centering on the impact of AI on journalism and I guess vice versa. So the question is, do AI companies need to pay for the training data that they use for their generative AI systems? So companies like OpenAI are, of course, using the output of journalists in part to train their models and so the question is, should they be paying those media outlets for using those works in AI projects? 

BP I'm going to step in here and give the company line which we've said in public, that if AI systems are training on data generated by humans, for example on Stack Overflow, we would hope that we can arrange some kind of partnership or licensing agreement. We said this in a blog many months ago that some of the resources that the AI companies are earning or have at their disposal will then go back into growing our knowledge community, or in the case of something else, supporting good journalism, let's say. I think that it's definitely an inflection point. It'll go to the courts, and I hope they get it right because I think they kind of got it wrong the last time around. It was a very painful transition for most news organizations as we entered the internet and the search age, and journalism is the fourth estate. It's important to democracy, it's important to civic health, so I hope we can get it right this time. 

RD The first draft of history.

BP The first draft of history. 

EM Well said. 

RD It's a toss up. You may need a sort of professional vetted content, that text in there that is sort of the best of the best, but on the other hand, if you're making money on a product that is essentially creating derivative works on everything, you probably should spread that money around and pay the people you're using the data of. But I am not a lawyer. 

BP This is not legal advice.

RD That's right. 

BP And unclear with private companies and other things, how much money or if any money is really being made in the sense of a profit so far in this world. 

EM For sure.

BP All right. We dove into this subject many times in the five years I've been here. 2019 to now there's been ups and downs. I do think, barring sort of the immediate, quick nosedive after the pandemic, the last year has seen quite a lot of layoffs across companies in the tech sector big and small. And I don't think– I'm making a 2024 prediction, unfortunately, that things are going to get better anytime soon, because the sort of stance on what kind of business people should be running has changed from ‘grow and it doesn't matter if you lose money’ to ‘you need to show us some profits,’ and that is not an easy transition to make. There were a bunch more layoffs this week at some very large startups that are still private, but among extremely well known, I won't name them here, and then hundreds of job cuts across engineering divisions that continue at large tech companies. And I read something the other day– Ryan, I could use your take on this, but it was that there's a generation of engineers– I don't know how you define generation, let's call it 10 years– who came out of college, got work, got great jobs, made gobs of money, moved around from maybe one to two to three places and have never experienced a downturn. This is their first downturn. So the feeling of insecurity as opposed to ‘I've always got 14 recruiters in my inbox on LinkedIn’ is a big shift I think for young engineers who didn't go through a dotcom cycle or a 2008 cycle. 

RD The folks who didn't go through that, although to be honest, this feels like a different set of layoffs. I remember the dotcom boom and bust cycle, but we didn't have huge companies at that point. We didn't have the really big tech companies doing layoffs, and it felt like the folks who were doing layoffs were these old legacy companies. Folks were like, “We're trying to adjust, trying to pivot,” and now the pivot has come for the tech companies. 

EM It's interesting actually to hear your perspective on what makes things feel differently this time around because I hadn't really thought about the fact that we have so many companies that are either well established but have that sort of startup energy laying a bunch of folks off and then actually startup companies, too, accounting for a big percentage of those layoffs. Something I heard from somebody else who's been on the tech job market lately is that it's no longer everybody just competing for the startup jobs or the big tech jobs. It's the same people who are competing at both ends of that spectrum, and perhaps that's also something that we've kind of not seen before in previous downturns like in 2008 or with the original dotcom bust.

BP Right.

RD And I wonder if these are still knock on effects of that section 124 or 174 tax code change we were talking about a few weeks ago where software engineering salaries now have to be amortized over five years and if now folks are thinking, “Oh, we cut these salaries. We can't just write them off. We can't just take them as expenses anymore, but we'll have these expenses to write off a long time in the future.”

BP I think that it sort of adds fuel to the fire. The change in interest rates played a big role and then that will only compound it by adding extra strain on startups.

RD And some of these are directly attributed to AI. I think there were some AI translators at a company to be named later that were laid off. 

BP Well, I said this back on the podcast many moons ago when Paul Ford and Sarah were my co-hosts and I talked about it with Cassidy and Ceora and here we are again. So condolences and thoughts and everything out to folks in the software industry who are affected, and if you want to discuss it, hit us up at the email or on Twitter and we'd be happy to share your story with the community or just talk about some of the issues you are interested in to help you process it or get it out there in the world. All right, I'll switch to something fun for our last little goodbye. We're going to share a great link in the show notes from the Blog of Dr. Moron –I hope that's his real name– who built a computer out of Lego, and it has a little switch that can flip and flop from one to zero, a reset, and a set for a clock, a NOR gate. And there's a line here at the end which I really love: “The point of this device, if there is one, is to separate out the concept and theory of computation from the digital electronic devices we most commonly associate with computing. There is nothing inherently electronic about computers. It just so happens that it's much easier to design computing devices using components driven from electricity, at least that is the state of the known inventions so far.” So I thought that was a fun little way to think about the world. 

RD I mean, computing is just a series of zeros and ones. Currently we store them with electrons and electricity, but if you have pins and holes, you can create a Turing complete computer.

BP Yeah, and as we mentioned before, quantum computers don't store them digitally anymore as a zero or a one. They store them as what? 

RD It's like a zero, a one, and both, I think. 

BP The qubit. 

RD Yeah, the qubit. It's interesting to see people be like, “Here's a computer in Lego or a Turing complete computer within Minecraft or something.” That's another one of those where they're going to create a Turing complete computer with, I think it was in The Three-Body Problem, the book. They used the position of people to create a Turing complete computer.

BP And a qubit is a physical thing. It's a device that behaves as a two-state system based on, let's say, the spin of the electrons. So it almost gets back to that idea that you can use the real world as a computer if you want to.

[music plays]

BP All right, everybody. That is your sci-fi far-out ending for this week. Let's get you a quick shout out to somebody great in the community. This badge was awarded one hour ago to sinθ. The question is, “How to use libraries.” Ooh, that's a bit broad, but it was asked 12 years ago, has been viewed 150,000 times, and is still clocking lots of views every single day. So congrats to sinθ for asking a great question. I'm sure many people, it seems, have had the same one. I am Ben Popper. I'm the Director of Content here at Stack Overflow. You can always find me on X @BenPopper. You can always hit us up, if you want to come on the program or pitch us something for the blog. We're all ears. And if you like the show, leave us a rating and a review. It helps. 

RD I'm Ryan Donovan. I edit the blog here at Stack Overflow. You can find the blog at, and you can reach out to me on X @RThorDonovan. 

EM My name is Eira May. I'm a Senior Writer at Stack Overflow also. You can find me on our blog too. You can also find me on Twitter @EiraMaybe. 

BP Thanks for listening, and we will talk to you soon.

[outro music plays]