Do you know what TCP/IP means? (Hint: You’re using it right now.) What about CDMA? Or GPT? While the concepts and execution of these technologies are clear to most of us who have been on the internet nearly our whole lives, the acronyms we use to define them are often inscrutable. On this week’s episode, we welcome WIRED’s AI reporter Will Knight onto the show. Along with our hosts Michael Calore and Lauren Goode, the trio takes turns quizzing each other on what exactly these acronyms stand for. Michael is asked to unpack various terms from the early internet era, Lauren is tested on acronyms from the mobile era, and Will tells us what all the AI-related abbreviations mean. Everyone does a pretty good job even if nobody earns a perfect score. Play along at home; maybe you can best our hosts with your arcane knowledge of internet minutiae.
Show Notes
Read Steven Levy’s story about the Google research paper that kickstarted the transformer-based AI boom.
Recommendations
Will recommends the book The Rise and Fall of the EAST by Yasheng Huang. (Watch their conversation at MIT’s Starr Forum.) Lauren recommends the Forest app for the Pomodoro work method. Mike recommends The Jargon File.
Will Knight can be found on social media @WillKnight. Lauren Goode is @LaurenGoode. Michael Calore is @snackfight. Bling the main hotline at @GadgetLab. The show is produced by Boone Ashworth (@booneashworth). Our theme music is by Solar Keys.
How to Listen
You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how:
If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts, and search for Gadget Lab. If you use Android, you can find us in the Google Podcasts app just by tapping here. We’re on Spotify too. And in case you really need it, here's the RSS feed.
Transcript
Note: This is an automated transcript, which may contain errors.
Lauren Goode: Mike.
Michael Calore: Lauren.
Lauren Goode: How long have you been covering the internet for now, as a journalist?
Michael Calore: Oh, over 20 years.
Lauren Goode: How long have you been at WIRED?
Michael Calore: Over 20 years.
Lauren Goode: Basically since the earliest days of the consumer internet.
Michael Calore: Yes. I've been online since I was a preteen. Sort of like that character in Almost Famous following around the band while I was a youngster.
Lauren Goode: Ah, don't make friends with rock stars.
Michael Calore: Yes, or the nerds.
Lauren Goode: How often would you say you still get tripped up on internet terms and protocols and acronyms?
Michael Calore: Oh, a lot. I mean, it's never ending, especially now with AI, which I don't really follow as a journalist, and even if I did, I don't think I could keep track of all the acronyms.
Lauren Goode: I agree, and I really think we need to demystify all of this for people, and what better way to do it than to make ourselves look like idiots.
Michael Calore: Yeah, OK. Sure. I think can think of better ways.
Lauren Goode: Let's do it.
Michael Calore: Let's do it.
[Gadget Lab intro theme music plays]
Lauren Goode: Hi, everyone. Welcome to Gadget Lab. I'm Lauren Goode. I'm a senior writer at WIRED.
Michael Calore: And I'm Michael Calore. I'm WIRED's director of consumer tech and culture.
Lauren Goode: We're also joined this week by senior writer Will Knight, who covers AI for us at WIRED. Hi, Will.
Will Knight: Hello there.
Lauren Goode: It's great to have you back on. So we're doing something a little bit different today. We are turning the Gadget Lab into a quiz show, but before you all turn to another podcast, seriously, stay with us, because we're going to attempt to define and explain all of the acronyms of the earliest and the most current consumer internet, stuff that you're hearing or maybe you even say all the time, like DARPA and TCP/IP and SMS and LLM. Do a shot every time you hear LLM on a WIRED podcast these days.
Now, this was partly inspired by an early 2000s book that I happened to reading recently. It's called Dot.Con by The New Yorker writer John Cassidy, and despite the fact that I know a little bit about the early internet, I mean, like Mike, I've been online since the mid-'90s. I was actually floored by how many terms I didn't know. And it just got me thinking, "We should do a podcast on this. Let's break this down into three parts. The early internet, the mobile era, and now the era of AI."
And we've brought in Will, because he is our expert AI reporter. So first, Will, you and I are going to quiz Mike, because he's the guy who's already established that he's been around forever, since the dinosaur age of the internet.
Michael Calore: Thank you.
Lauren Goode: You're welcome. And you have not seen these in advance.
Michael Calore: Nope.
Lauren Goode: Will and I have a shared doc with each other. We've crafted some ideas that we think some you're going to get easily and some, I think, might stump you.
Michael Calore: OK, I look forward to this. I'm not allowed to look them up while you're asking. Right?
Lauren Goode: No, you are not allowed to look them up. I was not allowed … none of us were. And the way we're going to break it down is after this you and Will are going to quiz me on the era of the mobile internet, because that's when I started covering tech. And then, finally, we're going to end with Will on AI. Will, would you like to go first in quizzing Mike on the wonky acronyms of the early internet?
Will Knight: Sure. I would be delighted. I want just to preface it by saying, I'm terrible at remembering acronyms in the best times and there are a billion out there in AI, so I'm going to do terribly. But the first—
Lauren Goode: Look at this—
Will Knight: … one—
Lauren Goode … caveating already.
Will Knight: The first one on our list is … it's not DARPA, it's ARPA, but they're related.
Michael Calore: ARPA. Oh, boy. Kicking it off with a bang here.
Lauren Goode: This predates the consumer internet.
Michael Calore: Let's see. DARPA, the D is defense, because the internet started as a defense project, a US government defense project. Right? I can't remember what comes after defense in DARPA, so I can't remember what ARPA is. Research project? Something something?
Lauren Goode: Mm-hmm, mm-hmm. Yeah, no. You're getting it.
Michael Calore: Is that right?
Lauren Goode: Yeah, yeah. I mean, if you think of the letter A and what the internet was at the time, it was pretty—
Michael Calore: Autonomous? No.
Lauren Goode: We're getting there.
Michael Calore: I don't know. It has something to do with—
Lauren Goode: It was the Defense Amazing Research Project.
Michael Calore: American?
Lauren Goode: No. No, it actually didn't start in America.
Michael Calore: OK, I—
Lauren Goode: Did it?
Michael Calore: I don't—
Lauren Goode: Well, the earliest internet did not.
Michael Calore: I give up. What does ARPA stand for?
Lauren Goode: Will, would you like to tell him?
Will Knight: It's Advanced Research Projects Agency.
Michael Calore: Advanced Research Projects Agency. OK. So was DARPA the Defense Advanced Research Projects Agency?
Will Knight: Yeah.
Lauren Goode: Yes.
Will Knight: That's right.
Lauren Goode: Mm-hmm, mm-hmm. This one I think you're going to get.
Michael Calore: OK.
Lauren Goode: BBS.
Michael Calore: BBS? Bulletin board system.
Lauren Goode: Yay.
Michael Calore: Yes, BBSes.
Lauren Goode: Good job.
Michael Calore: So a bulletin board system was like a computer in a basement somewhere or under someone's bed or in the closet in their mom's bedroom where they hosted a message board and you would call, on your modem, to the message board and leave messages, and then hang up and go about your day. And then, hours later, go back and read people's replies. It was a community forum.
Lauren Goode: Mm-hmm. Do you know where it started? This is a bonus question.
Michael Calore: Where BBSes started?
Lauren Goode: Mm-hmm.
Michael Calore: I would just assume the started in suburban North America.
Lauren Goode: Close. Well, it was Chicago in the 1970s during a blizzard.
Michael Calore: Oh, nice.
Lauren Goode: Yeah, two guys kind of patched it together.
Michael Calore: Nice.
Lauren Goode: Uh-huh.
Michael Calore: OK.
Will Knight: Wow. I was an avid bulletin boarder, I guess, when I was quite young. Yeah.
Lauren Goode: I was not.
Michael Calore: It's all there was.
Will Knight: Are they still around?
Michael Calore: Yes, I'm sure they are. I mean, it's basically Reddit, right? But it's all that was around. There was a whole lot you could do on the internet, other than go to BBSes and chat rooms.
Lauren Goode: Right. And Google, I think, acquired it. Am I remembering that correctly?
Michael Calore: They—
Lauren Goode: What was the—
Michael Calore: They acquired the big Use—
Will Knight: Usenet.
Lauren Goode: Oh, Usenet.
Michael Calore: Usenet.
Lauren Goode: Correct. They acquired Usenet.
Will Knight: Which is—
Lauren Goode: Yes, BBS was started by two guys in Chicago in the '70s. Usenet was actually started at Duke University, I believe.
Michael Calore: Yes.
Lauren Goode: Yeah. OK, fun one.
Will Knight: Simpler times.
Lauren Goode: Simpler times indeed. Mike, what about ABR?
Michael Calore: ABR?
Lauren Goode: Mm-hmm.
Michael Calore: ABR?
Lauren Goode: I didn't know this one. Will put this on there.
Michael Calore: I don't know what that means.
Will Knight: Did I?
Lauren Goode: I think you did, Will.
Will Knight: I can't have, because I don't know what that means.
Lauren Goode: Did I dream it up?
Michael Calore: I don't know what ABR is.
Lauren Goode: Available bit rate.
Michael Calore: Oh, available bit rate. OK, OK. So that's like when you call a server the maximum baud, the maximum BPS, baud, you can connect at, so 1,200, 2,400, 9,600. You're just staring at me.
Will Knight: I don't know.
Lauren Goode: I just felt like you'd go on.
Michael Calore: OK. These are good.
Will Knight: We don't use baud enough. I think—
Michael Calore: We don't.
Will Knight: I don't know what it means, but it's a great term.
Michael Calore: B-A-U-D, it's BPS, basically. It's the same. Bits per second. It's the same thing.
Lauren Goode: Right, yeah.
Will Knight: OK.
Lauren Goode: Yeah.
Will Knight: Makes me thinks of one of those modems making that noise.
Michael Calore: Yes, that's exactly what baud is.
Will Knight: Next one, TCP/IP.
Michael Calore: TCP/IP? Otherwise known to Americans as TCP/IP?
Will Knight: Oh, sorry. Yeah.
Michael Calore: TCP/IP. OK.
Lauren Goode: Stroke. I've never heard that before.
Michael Calore: That's how they say slash in Britishese.
Lauren Goode: OK. All right, then.
Michael Calore: Transferred content protocol/internet protocol.
Lauren Goode: Mm, close.
Michael Calore: Or did I miss the C?
Will Knight: Pretty close. No, you had control in there. I think you just had it all jumbled up, transmission control protocol/internet protocol.
Michael Calore: Transmission control protocol/internet protocol.
Will Knight: TCP. Yep.
Michael Calore: TCP/IP it's packets on the internet. We still use TCP/IP today. Right? Yes.
Lauren Goode: Correct.
Michael Calore: Internet traffic is TCP/IP traffic.
Lauren Goode: I just realized I'm not counting how many you're getting, but I think-
Michael Calore: I think I've gotten two.
Lauren Goode: Two out of four. OK.
Michael Calore: OK.
Lauren Goode: OK, next one.
Michael Calore: That's steaks and Cadillacs in the big leagues.
Lauren Goode: I admittedly didn't know that this was an acronym. BASIC.
Michael Calore: BASIC?
Lauren Goode: Mm-hmm.
Michael Calore: The programming language?
Lauren Goode: Mm-hmm.
Michael Calore: Oh, I don't know what it stands for.
Lauren Goode: Beginner's All-Purpose Symbolic Instruction Code.
Michael Calore: That sounds utilitarian and correct.
Lauren Goode: All right, two out of five.
Michael Calore: OK. How many do we do?
Lauren Goode: I don't know how many we're doing actually. This is all very organized.
Will Knight: GUI.
Michael Calore: GUI?
Will Knight: GUI. Yeah.
Michael Calore: Graphical user interface. GUI.
Will Knight: Correct.
Michael Calore: Yeah.
Lauren Goode: Correct.
Michael Calore: So that was a new thing when we moved away from text-based interfaces and we got a mouse and a pointer and icons and a desktop. That's a GUI. Yeah, I remember the birth of GUIs.
Lauren Goode: That's correct.
Michael Calore: Sweet.
Lauren Goode: This one comes directly from the book that I was reading. It's a little unusual. I would say it's not an internet protocol, but it's very related to the early 2000s internet. PCLN.
Michael Calore: PCLN?
Lauren Goode: Not an acronym, but it stands for something. I'm just breaking all the rules here.
Michael Calore: I don't know.
Lauren Goode: It's a company.
Michael Calore: It's a company. PCLN. OK, does this company make personal computers?
Lauren Goode: This company was one of the prime examples of boom and bust.
Michael Calore: Boom and bust. Is it Petco?
Lauren Goode: No. But it's—
Michael Calore: Oh, it is a stock ticker?
Lauren Goode: Mm-hmm.
Michael Calore: Ah, OK. It's a stock ticker for PCLN. It's not Compaq, it's not Gateway.
Lauren Goode: It starts with P.
Michael Calore: It starts with P? I don't know.
Lauren Goode: Priceline.
Michael Calore: Priceline.
Lauren Goode: Mm-hmm.
Michael Calore: Never would've gotten that.
Lauren Goode: I know. That's kind of a tough one.
Michael Calore: Yeah, that is a tough one.
Lauren Goode: Will, would you have gotten that?
Will Knight: No, not at all. What is Priceline?
Lauren Goode: Priceline was a travel website.
Michael Calore: Yeah, yeah.
Will Knight: Oh, OK. The thing with William Shatner.
Lauren Goode: With William Shatner. I should've given you that clue.
Will Knight: He built it, I believe.
Michael Calore: Yes.
Will Knight: Yeah.
Lauren Goode: Definitely was not just paid loads of money to promote it. He built it.
Michael Calore: OK, give me one more, because I'm deeply uncomfortable, and I want to end this on a pretty good batting average.
Lauren Goode: Oh, Will, do you want to do this next one? It's going to make you deeply uncomfortable.
Will Knight: OK. A slash S slash L. A/S/L
Michael Calore: OK. A stroke S stroke L? A/S/L. Is that what you're saying?
Lauren Goode: Oh my gosh. Stop saying stroke.
Will Knight: Yeah, more proper.
Michael Calore: OK. That's age/sex/location. Is that right?
Lauren Goode: That is correct.
Michael Calore: So if you were chatting with somebody and you usually would say A/S/L?, because you wanted to know their age, their sex, and their location. Sex, of course, meaning gender, and that was an AOL kind of chat thing that I did not really participate in as much, but I'm aware of that, because I was editing WIRED stories where people were referencing that acronym. That's how long I've been doing this. So yes—
Lauren Goode: That's pretty great.
Michael Calore: … that did make me deeply uncomfortable. Thank you.
Lauren Goode: Brought back memories. Huh?
Michael Calore: Yeah, it did.
Lauren Goode: It was just for finding friends.
Michael Calore: Sure, yeah.
Lauren Goode: Just like the rest of internet.
Michael Calore: Just like the rest of the internet. OK, so is this where I get to start asking you questions? And—
Lauren Goode: Yeah, how did you-
Michael Calore: … Will and I—
Lauren Goode: … score? I really wasn't keeping track. I think you got four out of—
Michael Calore: I think I got all of them right.
Lauren Goode: Let's go with that.
Will Knight: I think so.
Lauren Goode: Three, four, five, six, seven, eight, nine. We gave you eight, and I think you got about half.
Michael Calore: OK.
Lauren Goode: I don't know what the prize is, but congratulations.
Michael Calore: The prize is bragging rights, as always.
Lauren Goode: OK.
Michael Calore: All right. OK, Will, we're going to quiz Lauren now. And so, we're moving on—
Will Knight: All right.
Michael Calore: … to the mobile era, which I guess is roughly the turn of the century to about two years ago or 2005? '06?
Lauren Goode: I would classify it as 2007, when the iPhone was launched and then the app store the following year, 2008 onward.
Michael Calore: OK, OK. OK, we'll do that.
Will Knight: Aren't we still in the mobile era? No, we're in the—
Lauren Goode: I think so.
Will Knight: … metaverse era.
Michael Calore: Yes. We've gone full meta. All right, well, so here we go, for the mobile era.
Will Knight: I don't know how big this was in America, but WAP, W-A-P, do you know this?
Michael Calore: Not the song, Lauren. Not the song.
Lauren Goode: All right, OK. No, I'm going to make this up. Wireless access protocol.
Michael Calore: That's like—
Will Knight: Close enough, close enough.
Michael Calore: … very, very close.
Lauren Goode: OK. What is it?
Will Knight: Wireless application protocol.
Lauren Goode: Oh.
Will Knight: I don't know that this is where they were like, "We're going to reinvent the web for the mobile era and it'll be really terrible and clunky," and just little pixelated websites. It was kind of a thing on Nokia phones in Europe for about a year.
Michael Calore: We had that here, too.
Will Knight: OK.
Lauren Goode: Yeah, we did.
Michael Calore: Yeah. The first mobile browsers were WAP browsers.
Lauren Goode: I don't think I realized they were called that.
Michael Calore: Yeah. Safari, I think, was the first mobile browser that was an actual browser.
Lauren Goode: Huh.
Michael Calore: Yeah.
Lauren Goode: One note about this era, I think that if I'm going to get something confused quite a bit, it's service and system, because a lot of the S's can refer to either in the mobile era.
Michael Calore: Oh, there's a lot of—
Lauren Goode: But it's going to—
Michael Calore: … S's that mean neither of those in this list that I'm looking at-
Lauren Goode: Oh, boy.
Michael Calore: ... right now.
Lauren Goode: OK, OK.
Michael Calore: OK, second one. MVNO.
Lauren Goode: Oh, oh, oh. Mobile video network operator.
Michael Calore: Close.
Lauren Goode: No, mobile … it's not video, is it?
Michael Calore: Mm-mm.
Lauren Goode: It's mobile … I'm just going to make up something. Mobile vector network operator, that's a word people use a lot.
Michael Calore: Virtual.
Lauren Goode: Virtual. Yes.
Michael Calore: Mobile virtual—
Lauren Goode: Yes, I—
Michael Calore: … network operator.
Lauren Goode: … do remember that.
Michael Calore: So what is it? Can you define it?
Lauren Goode: No. I don't remember what it is.
Michael Calore: It's like when a company leases spectrum from somebody who owns a network.
Lauren Goode: Yes, yes. It's all coming back. This is early mobile.
Michael Calore: Yes.
Lauren Goode: I was thinking app era. OK, this is fun.
Will Knight: OK. Next one, SIM or S-I-M.
Lauren Goode: Oh my god.
Michael Calore: This is actually hard.
Will Knight: Or eSIM.
Lauren Goode: I think I'm going to quit while I'm ahead and just quit now. Well, eSIM is electronic SIM, but SIM is … You know who I keep thinking of right now who would know all of this is Dieter Bohn.
Michael Calore: Yeah, probably.
Lauren Goode: I wish I had a lifeline to Dieter.
Michael Calore: So SIM is a really hard one, because I never would've guessed that it stands for what it stands for.
Will Knight: I wouldn't.
Lauren Goode: OK, so it is the thing that gives you connectivity on your mobile device, so I'm going to go with satellite. Is that correct?
Michael Calore: Mm-mm.
Lauren Goode: OK. System?
Michael Calore: Nope.
Lauren Goode: OK. Simulated.
Michael Calore: No. I'm telling you, it's very esoteric.
Will Knight: Correct.
Lauren Goode: OK. I don't know, I don't know.
Michael Calore: It's the Subscriber Identity Module.
Lauren Goode: Oh, come on.
Michael Calore: Yeah, isn't that weird?
Lauren Goode: Who came up with that?
Michael Calore: I don't know.
Will Knight: It is weird. That is very weird.
Lauren Goode: Oh, wow. OK.
Will Knight: I mean, there probably is a simulated identity module somewhere. You can just say, "I was thinking of that one."
Lauren Goode: Hmm, simulated. OK. Wow, I think I'm 0 for three, 0 for four?
Michael Calore: We'll get you some good ones.
Lauren Goode: OK.
Will Knight: OK. Well, all right, a softball. SMS.
Lauren Goode: Short Messaging Service.
Michael Calore: Yeah.
Will Knight: It's actually simulated message subscriber.
Michael Calore: OK. Related question—
Lauren Goode: MMS.
Michael Calore: No.
Lauren Goode: Darn.
Michael Calore: RCS.
Lauren Goode: RCS. This is one of the ones where I'm going to get system and service confused. It's Rich Communications Service?
Michael Calore: Yes.
Lauren Goode: OK, great. Yeah. So that there's SMS, which is Short Messaging Service, which is text based and happens over the wireless networks. And then, MMS is when SMS basically got upgraded to multimedia service. And then, now, RCS is the wireless network-backed and Google-backed sort of new era of MMS that's bringing richer communications to what would typically be text-based messaging.
Michael Calore: Yes.
Lauren Goode: Yeah.
Michael Calore: Tapbacks, stickers.
Lauren Goode: Yeah, all the fun stuff. All the—
Michael Calore: All the fun stuff.
Lauren Goode: … ways we communicate. All the ways we have entire relationships now through tapbacks. Thanks for those easy ones, guys. I feel great.
Michael Calore: OK, Will, pick another one.
Will Knight: OK. BBM.
Lauren Goode: BlackBerry Messenger, which predates iMessage as one of the first peer-to-peer direct messaging services on your phone that was actually owned and operated by the BlackBerry network.
Michael Calore: Hey, now.
Lauren Goode: That was really fun. I miss BBM.
Michael Calore: My favorite BBM thing is when they started buying product placement in television shows, so characters would say—
Lauren Goode: "BBM me."
Michael Calore: … "Oh, I'll BBM you." And then, they'd pull out their BlackBerry-
Lauren Goode: Oh, it's so great.
Michael Calore: … and the camera would show them on their BlackBerry doing that and it was terrible. OK, here's another one for you. SoC. Sometimes—
Lauren Goode: System-on-a-chip.
Michael Calore: Yes.
Lauren Goode: Yes. Yeah, that refers to when it's one piece of silicon, but it's put together to create a system, and so you might have one core or basically one chip in the system that's dedicated to ML, which is machine learning, or you might have another one that's basically dedicated to the core processing. But yeah, you put them all together as a system, Qualcomm is a very well-known maker of system-on-a-chip. Apple makes its own now. Lots of people make them.
Michael Calore: OK, Will, throw her another one.
Will Knight: OK. How about CDMA?
Lauren Goode: This one is really tough.
Will Knight: Yeah. Was it like a precursor to 3G and it didn't last that long or something?
Lauren Goode: No—
Michael Calore: It was—
Lauren Goode: … it was a type of network. So early on, mobile phones were pretty much divided between GSM and CDMA. GSM was more popular in Europe. So Will, you might've been on GSM. Here, it was... Yeah, basically when you bought a mobile device, you had to specify, based on which wireless carrier you were on, what kind of device you were—
Will Knight: Oh, right. That's right, because—
Lauren Goode: … using.
Will Knight: … you had different ones. Yeah.
Lauren Goode: Yeah, and—
Michael Calore: Verizon was our CDMA network, I think.
Lauren Goode: That's right, that's right. And 3G was more in the category of LTE, which stands for long-term evolution, of the type of service you would get. But OK, I'm-
Michael Calore: You should've waited for us to ask you that, so you could get another point on the board.
Lauren Goode: OK. Shoot, OK. CDMA. It's consolidated?
Michael Calore: Nope.
Lauren Goode: Concentrated? Communication? Oh, man. Can you give me the first word?
Michael Calore: Code.
Lauren Goode: Code demystifier. Code debugging.
Michael Calore: OK.
Will Knight: This one is horrible. Sorry.
Lauren Goode: Code … All right—
Michael Calore: This one is really hard.
Lauren Goode: … what is it? What is it?
Will Knight: Code Division Multiple Access.
Lauren Goode: I would like to talk to the manager. What?
Will Knight: It's just so obvious. It's so logical. What are you talking about?
Lauren Goode: This is really fun, though. OK. Bring it on. What's another one?
Will Knight: I don't know if this is really mobile era, but it's three emojis. And eye emoji, lips emoji, and another eye emoji.
Lauren Goode: What?
Michael Calore: An eye emoji.
Lauren Goode: Yep, yep. An eyeball, uh-huh.
Michael Calore: And then, the mouth.
Lauren Goode: And then, the mouth.
Michael Calore: And then, an eyeball.
Lauren Goode: Oh.
Will Knight: This is actually a way more recent thing, think.
Lauren Goode: OK, so it's eye. Hear. No, that's an ear. I talk … I have no … No one's ever sent this to me. Am I being left out? What is this? Eye—
Will Knight: It stands for, "It is what it is."
Michael Calore: Mm-hmm.
Lauren Goode: Really?
Will Knight: Yeah. It's kind of a TikTok thing, I think.
Lauren Goode: Oh, wow. That's—
Will Knight: Sort of as a—
Lauren Goode: I have to say, I was picturing a totally different era here.
Michael Calore: Well, this was the very beginning of the pandemic this ramped up.
Lauren Goode: Oh, I mean, no, what I mean is you went really early mobile and then you fast-forwarded into the future. I was waiting for like ARPU and ATT and GPS and—
Michael Calore: Oh, GPS—
Lauren Goode: … stuff like that.
Michael Calore: … would've been a good one.
Lauren Goode: Yeah.
Michael Calore: Yeah, sorry.
Lauren Goode: Average revenue—
Will Knight: So maybe we should've had—
Lauren Goode: … per unit. App tracking transparency. I was clearly in kind of app mode. All right, though, you know what? I had a lot of learn.
Michael Calore: I think you did just fine.
Will Knight: We clearly need some Gen Z person to come on and do TikTok, incomprehensible-
Lauren Goode: We'd all fail.
Will Knight: ... TikTok slang.
Lauren Goode: Yeah, no. That would be the end of us.
Will Knight: We wouldn't have a clue.
Lauren Goode: All right. So Mike, I think that you're in the lead, technically. Whatever this competition is.
Michael Calore: I wasn't keeping score, but OK, thank you for keeping score.
Lauren Goode: Yeah, OK. This is really fun. Thank you for schooling me guys. We are going to take a quick break, and when we come back, we're going to spend the entire next segment talking to Will, quizzing Will about AI, because it's what everyone's talking about. So stay tuned.
[Break]
Lauren Goode: OK, so now we've brushed up on basically 25 years of tech acronyms. Congratulations. You're now everyone's favorite dinner party guest. It's time for AI. This is the part I think everyone is going to be most interested in right now, because you're hearing these phrases literally all the time, and some of them are probably hard to grok. See what I did that?
Michael Calore: Mm-hmm.
Lauren Goode: That's Elon Musk's AI thing.
Will Knight: That's not an acronym, is it?
Lauren Goode: No, it's not an acronym. But we couldn't—
Will Knight: Thank god.
Lauren Goode: … get through an entire show without mentioning Elon Musk. Mike, do you want to go first in grilling Will?
Michael Calore: Sure, OK. Will, I'm going to kick off with the hardest one on the list. LLM.
Will Knight: OK. That's large language model, which is what everything's built on these days, all those chatbots, and half of the internet built on large language models. Yeah, what to say about LLMs? Yes, it is an acronym that you're hearing all the time.
Lauren Goode: So if someone is at a dinner party and someone starts talking about LLMs in the context of AI, how would you describe, "Here was this era of AI, but now everyone's talking about LLMs because they do X."
Will Knight: OK, yeah. So language models are this example of generative AI and generative AI has been around for decades and decades, in that you generate stuff with an algorithm, but the large language model era is this era where they've figured out feeding it enough information, feeding it basically the whole of the internet and as many books, a lot of copyrighted material, as well, it turns out, into these particular types of algorithms, enables them to kind of conjure up very convincing seeming text. But yeah, the large is the key word there, because they're absolutely enormous algorithms and also fed huge amounts of information or data.
Michael Calore: Nice.
Lauren Goode: Nice. Thanks, Will. And that's our show. That's all you needed to know. No, there's another one that we're hearing a lot these days. GPT.
Will Knight: GPT. Oh, OK. Generative Pre-trained Transformer. Right?
Michael Calore: Yes.
Lauren Goode: Yay.
Will Knight: So that's what I'm talking about, the generative models. It's the type of AI algorithm that doesn't discriminate, doesn't recognize things in text or images, but generates stuff when given a prompt and pre-trained means that they are pre-trained, actually, on specific types of data before being trained on the entire internet and transformer is a certain type of algorithm. It's a neural network that can focus on lots of different stuff at once, which turns out is very useful for language, because you kind of need to know the end of a sentence and the beginning, to make sense of what's in the middle of it, sort it.
Steven wrote a great story about the paper, Attention is All You Need, which came out of Google, which sort of transformed AI by revealing you can do amazing things with language using these models the first time or it pointed that way.
Michael Calore: That's our colleague Steven Levy who wrote that.
Will Knight: Sorry, yes.
Michael Calore: We'll put a link in the show notes to it.
Lauren Goode: Yeah, you can read it now. It's really good. And a lot of people probably know this, but for those who don't, when you refer to something like ChatGPT, which was released by the company OpenAI, the T in ChatGPT comes from that Google paper, that group of Google researchers. So it's kind of a derivative of Google's tech.
Michael Calore: Yeah, OK.
Will Knight: Yeah, and this is in Steven's great story, but pretty much all of the people in that paper, the transformer paper, ended up creating their own … well, a lot of them went off and created their own startups and got billions of dollars of funding.
Michael Calore: That's why we all do it, though, right? To make billions of dollars.
Will Knight: Well—
Lauren Goode: Humanity be damned.
Will Knight: … speak for yourself. Yeah.
Michael Calore: OK. Here's another one for you, the third one. You're two for two, by the way. Congratulations. You're already whooping us.
Lauren Goode: He's going to sweep.
Michael Calore: NLP.
Will Knight: OK. That's natural language processing.
Michael Calore: Yes.
Will Knight: I think.
Lauren Goode: Mm-hmm. Pardon me, Will, it's processing.
Will Knight: Processing, sorry.
Lauren Goode: We're going to minus—
Will Knight: As opposed to—
Lauren Goode: … one point.
Will Knight: … NLU, natural language understanding, so that means doing stuff with language, not just understanding or not comprehending the language but doing the synthesizing sort of stuff.
Michael Calore: OK, so natural language understanding is so that a computer can understand what you're saying when you say something to it, but natural language processing, so that it can answer you in a way that it is humanish. Is that right?
Will Knight: Yeah, I think it just—
Michael Calore: Or conversational?
Will Knight: … refers to the bigger field of doing stuff with language, I think. Yeah, I think it can encompass voice recognition and processing, that sort of stuff. I think. I could be wrong.
Lauren Goode: We're going to go with you're right. Yeah. Here's the next one, which I think you'll get. RL.
Will Knight: Oh, reinforcement learning.
Lauren Goode: Mm-hmm.
Will Knight: So this is—
Lauren Goode: He's crushing it.
Will Knight: Well, these are all more recent, and I think you're being kind to me, because there are a billion abstract acronyms out there. RL is the type of algorithm that AlphaGo was, so it's quite different to things like ChatGPT in that it's the idea of having a computer learn through experimentation how to solve a particular task. So DeepMind didn't invent reinforcement learning, that was this guy called, I think it was Rich Sutton or he was a pioneer of it. But they figured out you could, with more powerful computers, get machines to do quite impressive tasks that it's impossible to program a computer to do.
So they started off with having it play Atari games better than person, and then famously demonstrated it on chess, the board game, which is very, very difficult to learn and play and it's very difficult to describe what makes a good move, so it's hard to break up. But the idea is it has reinforcement in the form of positive or negative feedback for a good or a bad move, moving towards a goal. Interestingly, the big thing that people are trying to do now, to move beyond ChatGPT is combine some of these two things. A lot of places are trying to do that.
So ChatGPT sort of goes off the rails and says mad things and it doesn't understand what numbers are, which is kind of a problem in an intelligence. But you can use reinforcement learning to have it, possibly, figure out how to perform specific tasks, maybe including things like math, but it's not demonstrated that it will work, but this is sort of an idea to take these two big things in AI and combine them.
Michael Calore: OK. Here's another one for you. LSTM.
Will Knight: OK. I have to mention the name of the guy who invented this, because he famously gets upset when people don't, it's Juergen Schmidhuber invented the LSTM. So—
Lauren Goode: God bless you.
Will Knight: Yeah. He's quite a character. Long short … shoot, something memory. Long short term memory?
Michael Calore: Yes.
Lauren Goode: Yep, correct.
Michael Calore: Oh, boy.
Lauren Goode: Yeah, I know. We should've made this harder. Mm-hmm.
Will Knight: So that's a type of neural network that can tap into memory, essentially, which is something they don't have. There are a lot of these different architectures. This is one that was quite old. It sort of predates a lot of deep learning stuff, but it was very important and influential, hence why Juergen likes to be credited with stuff. I spoke to him recently, and he was doing some very interesting things, trying to build these models that kind of argue with each other in order to figure out a task, like a kind of Gestalt thing that, if one isn't good at it, then the other one can figure it out, so you have these specialized networks.
Michael Calore: Wow.
Will Knight: I don't know. He's …
Michael Calore: Is he a big Stanislaw Lem fan?
Will Knight: He is a huge science fiction fan, actually. Why is … Oh, because of—
Michael Calore: The computers arguing with each other and competing—
Will Knight: Oh, of course. Yeah.
Michael Calore: … to try and figure out problems.
Will Knight: Yeah. I didn't think of that. Yeah, he's a huge sci-fi nerd. That must be, probably, his inspiration, perhaps.
Lauren Goode: So what you're describing is a new era of the virtual assistant remembering what you've talked about before. Because there was a version of this on Google Home Assistants several years back, at this point, I want to say at least five years back, where you would say to the Google Home or your assistant on your phone, "Hey, Google," ask it a question like, "How tall is LeBron James?" and then, without having to prompt it again, saying, "OK, and what team does he play for?" and having this volley back and forth where it had a limited amount of memory to remember what you asked initially. But now, Will, it's this movement towards you could ask ChatGPT for an itinerary for Barcelona and then come back days later and pick up the thread and just—
Will Knight: Oh, yeah.
Lauren Goode: … say something like, "What else should I add to my trip?" and it's going to remember what it was you talked about. And people are talking about this, too, not only in the form of customer service agents and stuff like that, but even in EQ AIs, ones that are meant to do more emotional tasks, if you're using one for therapy. Imagine coming back a week later and having it remember what you talked about before.
Michael Calore: Yeah.
Will Knight: Yeah. Because ChatGPT doesn't remember anything beyond the previous prompts, typically. I think they've added some more memory, but that idea of—
Lauren Goode: They have.
Will Knight: … having a bigger memory. OK, it's a big thing. But the other thing that he is looking at is the idea that … because you can have a small model that's better than GPT-4, if it's very much trained on a specific task. So it's the idea that you have a bunch of these ones that interact with each other, and then they can work as well as a really big one, but also, I think they're back and forth. The idea is that it kind of shakes out more … something cleverer. I don't know if it works.
Lauren Goode: OK. The next one, PaLM, which I'm pronouncing like a word, but it's P, lowercase A, and then capital L-M.
Will Knight: Oh, god. I'm not going to get this one.
Michael Calore: Yeah, we stumped him.
Lauren Goode: Yes.
Will Knight: I'm going to make something up.
Lauren Goode: I wouldn't get the first word.
Will Knight: It's language model at the end, probably.
Michael Calore: Oh, yeah.
Lauren Goode: Will, just so you know, before this episode, we joked about calling this WILLM, which would be W-I-L-L-M, the Will learning large language model.
Will Knight: One day my language model will be able to come in my place and get it all right, because it can look it up. But I don't know, pre-trained amazing language model.
Michael Calore: Nope.
Lauren Goode: Good guess. It's Pathways Language Model.
Will Knight: Oh.
Lauren Goode: Mm-hmm.
Will Knight: OK.
Lauren Goode: Yep. Yes, we got him.
Michael Calore: So who does—
Will Knight: So this was the—
Lauren Goode: This was Google.
Will Knight: … precursor to ChatGPT, right? They built some pretty amazing chatbots on top of it, but never released them, I think.
Michael Calore: OK.
Will Knight: I think that's right.
Michael Calore: Well, here's another one with a lowercase in it. This one is LLaMA. L-L, lowercase A, capital M, capital A.
Will Knight: Mm, I don't know this one either.
Lauren Goode: I think you do.
Will Knight: LLaMA.
Lauren Goode: Only because, I'm going to give you a hint, because I think you know the company behind it and that letter factors into this.
Will Knight: What some letters behind LLaMA?
Lauren Goode: Mm-hmm.
Will Knight: Two Ls, language … How does Meta? I don't know. I have no idea.
Michael Calore: It's Large Language Model Meta AI. LLaMA.
Will Knight: Makes perfect sense.
Michael Calore: Yeah, right? So there's extra letters in there. It's Large Language, lower case A, Model Meta AI, but there's not two Ms, it's just LLaMA, like the animal.
Will Knight: I should've at least got large language and made the rest up.
Lauren Goode: They really just shoe horned Meta in there.
Michael Calore: They did, yeah.
Lauren Goode: Yeah.
Will Knight: Because there was an alpaca, I think. I think they wanted to get something, I think, I might be hallucinating that.
Lauren Goode: Fitting.
Will Knight: Anyway.
Lauren Goode: The next one, which will be the last one, it's a bit of a trick, because it's not an acronym, but maybe you can describe where it comes from. DALL-E. D-A-L-L, dash, E.
Will Knight: Oh, I do know this. I might get this word wrong. Is it a portmanteau? Is that they way you say it?
Michael Calore: Yes, portmanteau.
Will Knight: It's a portmanteau, sorry, of Salvador Dali and WALL-E.
Lauren Goode: Good job.
Michael Calore: Excellent.
Lauren Goode: Wow.
Will Knight: I just remember that, because every story had to explain that it was portmanteau or whatever that word is. And so, I was like, "I better learn that word," which I haven't done.
Lauren Goode: And this is an image generator.
Will Knight: Right, right. This is the OpenAI image generator that is trained on lots of nice artwork. Yes, trained on the whole of the internet, which includes lots of artwork, I guess.
Lauren Goode: All belonging to OpenAI.
Michael Calore: Yes.
Lauren Goode: Fully within copyright.
Michael Calore: Yes, no issues there.
Will Knight: Yeah.
Lauren Goode: I think Will won.
Will Knight: I-
Lauren Goode: Handily.
Will Knight: Did I?
Michael Calore: I think so. Yeah, you got six out of eight.
Will Knight: Oh, OK. I felt like I fell off a cliff.
Michael Calore: Well, yeah, but you win all your games in the first half of a season, it's OK to lose some in the second half of the season.
Lauren Goode: Is that how it works? Don't you get closer to the playoffs in the second half?
Michael Calore: Yeah, but your stands are full all year round, if you win all your games in the first half of the season.
Lauren Goode: Spoken like a true capitalist.
Michael Calore: I'm an A's fan.
Lauren Goode: He is. He's wearing an A's hat right now.
Will Knight: But it might be slightly different if you guys were writing about CDMA and the like every week.
Michael Calore: That's true.
Lauren Goode: Yeah, that's true. This is true. I mean, I referenced those things a lot. When I first started writing about tech I was a video journalist at The Journal, and then I started writing and I remember my first story, it was a MDTV, mobile DTV. But then, after that, I just sort of cruised right into modern smartphones and app store. So the early stuff, that was really fun.
Michael Calore: Foundational.
Lauren Goode: Super fun to learn about.
Michael Calore: You did well.
Lauren Goode: I don't know. Not so much. I can admit defeat. What does Will get? What's his prize?
Michael Calore: He gets to go first on recommendations.
Lauren Goode: Ooh, good one. Let's take a break and come back with those.
[Break]
Lauren Goode: Will, as the winner of our acronym competition, all we can give you is the option to go first in recommendations, so have at it.
Will Knight: I wish I'd known, because I would've deliberately failed, to buy myself more time, because I'm literally trying to come up with this on the fly, but I did think of it before. So I'm going to recommend a book that I read recently, I'm holding it up for the listeners on Zoom, The Rise and the Fall of the EAST by Yasheng Huang. The EAST in the title stands for Exams, Autocracy, Stability, and Technology. It's a kind of incredibly in-depth historical analysis or Chinese bureaucracy, which sounds like a real page-turner, but it's actually incredibly well done and does a lot to actually explain how China got to where it is and why it's sort of this paradoxical place where it seems very innovative, but in some ways it's not as innovative or it's sort of held back and I found it very convincing as to why, also, China is where it is right now with the current leadership, and it's really fascinating, if you're interested in China and its technology.
Lauren Goode: And how did you hear about this book?
Will Knight: So I know Yasheng Huang from MIT, and I know his work. He's quite a famous scholar on China and its tech industry and they asked me to interview him on stage there, where I live down the road from MIT. So I had to furiously read it in two weeks, which was kind of intense, but I knew he had one in the works, because I'd been talking to him.
Lauren Goode: This is how modest Will is. I think a lot of us would lead with, "So I spoke to this really smart thinker and author recently and our talk was so scintillating and also he has a book." And Will kind of led with, "I'm reading this book." "And how do you know him?" "Oh, at MIT I happened to interview him."
Will Knight: I'm just very incompetent at presenting things, selling myself.
Lauren Goode: Not at all. Will, we will not only link to the book in the show notes, but we'll link to your talk, as well.
Will Knight: Thank you.
Lauren Goode: Mike, what's your recommendation?
Michael Calore: My recommendation is on topic for this week's show. It's called The Jargon File, and it's this open source text document that lives out on the internet that you can print out or you can just look at, it can be copied and pasted in other places. But it's basically a big glossary of hacker slang and acronyms and fun terms. There's a lot of humor in it, and it's a fun way to look at computer history through the language and the rhetoric that people have associated with computers and how hackers talk to each other. So yeah, The Jargon File is basically just like an A to Z dictionary and you look up words.
For example, some words that are in there, I'm just going to flip through. Bit rot is a good one. Cyberspace and cyberpunk are both defined in here. Dancing frog, kill file, the pumpkin, the patch pumpkin, the pumpkin holder, spoiler space, vaporware is defined in here. Windowz with a Z is defined in here, so it's a really fun way of looking into hacker culture and computer science over the years.
Lauren Goode: Cool. Who created it?
Michael Calore: It's an open source project, so a bunch of people have edited it and contributed to it over the years.
Lauren Goode: But someone must've started it.
Michael Calore: I'm sure somebody did start it, yes. I don't know who that person is. But it's very easy to find, because it's hosted multiple places and you just need to search for Jargon File. You search for those two words together and you'll find it. It's just a big HTML document.
Lauren Goode: Super cool, super nerdy. That's what we're here for.
Michael Calore: It's a lot of fun.
Lauren Goode: That's what we should rename our podcast.
Michael Calore: Super Nerdy?
Lauren Goode: Super Nerdy.
Michael Calore: Then we actually have to live up to that descriptor and I—
Lauren Goode: I know. It's true.
Michael Calore: … don't know, as a couple of English majors.
Lauren Goode: Oof, yeah. And sometimes my recommendations are very basic.
Michael Calore: OK, I want to hear yours. Is this all caps basic, B-A-S-I-C basic?
Lauren Goode: No, but I like how you're bringing it back. My recommendation: it's not at all academic or nerdy, it's an app called Forest. Some of you might be familiar with the Pomodoro method of working, which means you set a timer and you work for 25 minutes straight, no distractions. There are a lot of different apps that sort of take advantage of the Pomodoro method, and then create different user interfaces for it and different mechanisms. There are also web versions for people who get really distracted on their desktop with stuff flying into their browsers and 18 different browser tabs open.
This one is a mobile app. I think it cost me $2.99 to download. I just clicked a button and showed my face and I had it, but I'm pretty sure it was $2.99, because that's how we buy things on the internet now. And it simulates planting trees, so every time you work 25 minutes uninterrupted, it sets the timer, you've planted a tree at the end of it, so it's the Pomodoro method a little bit gamified, and it's just a good way to cut down on distractions, if you feel like you're a little bit ADHD as you're working. Yeah, so check out the Forest app. And then, if you want to work for an extended clip, you can just set 25-minute increments with a five-minute break in between. I'm not using it quite like that yet, I'm just picking a 25-minute block and going for it.
Michael Calore: That's nice.
Lauren Goode: Helpful for writers.
Michael Calore: Yes.
Lauren Goode: For sure.
Michael Calore: Or anything task-based.
Lauren Goode: Correct. Yeah, all of a sudden thinking about doing administrative work or filing expenses or something like that becomes a lot more tolerable, if you just think, "I can do this for 25 minutes."
Michael Calore: Go make a tree.
Lauren Goode: Go plant a tree, a little virtual tree.
Michael Calore: That's lovely.
Lauren Goode: In a simulated world, like SIM. I had to bring it back, too. Before we say goodbye, I just want to give a shout out to some folks who have been leaving us reviews on the Apple Podcasts app. We love the reviews. We genuinely read them. We appreciate the feedback and thank you. Yeah, for everyone who's listened to this point, go leave us a review.
Michael Calore: We could say thank you to the people who've left reviews on the Google Podcast app, but that's gone, so there isn't one anymore.
Lauren Goode: RIP Google Podcasts. I know, where else do people leave podcast reviews?
Michael Calore: YouTube probably.
Lauren Goode: Oh, all right.
Michael Calore: We'll see.
Lauren Goode: We're not on there.
Michael Calore: Sure we are.
Lauren Goode: We are?
Michael Calore: Our feed publishes there, I think, but you don't see our faces. You just see the robot blowing the bubble.
Lauren Goode: Yeah. I love—
Will Knight: Are you guys—
Lauren Goode: … our robot.
Will Knight: … on TikTok? Do you have a TikTok?
Lauren Goode: Sometimes. Our podcast is not on TikTok. I go on. Mike's on TikTok a lot. He needs the—
Michael Calore: All the time.
Lauren Goode: … Pomodoro app to focus, because TikTok is ruining his life. He loves the dancing teen videos. It's really weird. He's shaking his head right now.
Michael Calore: Because I don't. I don't.
Lauren Goode: You know what the TikTok emoji were.
Michael Calore: Well, yeah, because I had the sheet that Will and I shared.
Lauren Goode: Oh, right. OK. Speaking of, Will, thank you so much for joining us.
Will Knight: Thank you for having me, and for letting me win. I appreciate it very much.
Lauren Goode: Very welcome. It's been an absolute pleasure. And Mike, thanks for being a great cohost.
Michael Calore: Of course, anytime.
Lauren Goode: And thanks to all of you for listening. If you have feedback you can find us all on the social networks, just check the show notes. Our producer is the excellent BA, otherwise known as Boone Ashworth. Goodbye for now. We'll be back next week.
[Gadget Lab outro theme music plays]
