[00:15.010 --> 00:16.250] [SPEAKER_00]: Got it.
[00:17.350 --> 00:22.810] [SPEAKER_00]: You are listening to radio station WBAI.
[00:23.070 --> 00:26.110] [SPEAKER_00]: It's 99.5 FM here in New York.
[00:26.550 --> 00:27.770] [SPEAKER_00]: WBAI.org.
[00:28.270 --> 00:29.530] [SPEAKER_00]: Everywhere, all over the place.
[00:30.390 --> 00:36.870] [SPEAKER_00]: Time is just about 7 o'clock, and it is time for another exciting edition of Off The Hook.
[01:17.630 --> 01:20.950] [SPEAKER_00]: A very good evening to all of you.
[01:21.630 --> 01:23.470] [SPEAKER_00]: The program is Off The Hook.
[01:24.230 --> 01:24.870] [SPEAKER_00]: Rob T.
[01:25.010 --> 01:29.350] [SPEAKER_00]: Firefly here with you, and we are back after a week off the air.
[01:29.350 --> 01:32.830] [SPEAKER_00]: And I am here, joined today by Gila.
[01:33.030 --> 01:33.430] [SPEAKER_04]: Good evening.
[01:34.890 --> 01:35.890] [SPEAKER_00]: Let's try that again.
[01:36.750 --> 01:37.490] [SPEAKER_00]: Hi, Gila.
[01:37.730 --> 01:37.930] [SPEAKER_04]: Hi.
[01:38.230 --> 01:39.030] [SPEAKER_04]: Can you hear me now?
[01:39.270 --> 01:40.870] [SPEAKER_00]: I can hear you a little better now.
[01:40.990 --> 01:41.690] [SPEAKER_00]: We'll keep working on that.
[01:41.930 --> 01:44.470] [SPEAKER_00]: And out in Signal-land, we have got Alex.
[01:45.750 --> 01:46.590] [SPEAKER_01]: Yes, you do.
[01:46.830 --> 01:47.510] [SPEAKER_01]: I'm here.
[01:48.430 --> 01:48.910] [SPEAKER_00]: All right.
[01:49.810 --> 01:54.970] [SPEAKER_00]: Now we're stabilizing some levels and putting some stuff out there onto the airwaves.
[01:55.530 --> 01:57.890] [SPEAKER_00]: We were not on last week.
[01:58.790 --> 02:03.370] [SPEAKER_00]: We had planned to be on, but those plans changed, and...
[02:03.370 --> 02:04.250] [SPEAKER_04]: It's my fault.
[02:04.250 --> 02:04.710] [SPEAKER_04]: I own it.
[02:04.990 --> 02:07.410] [SPEAKER_04]: It's completely... It is 100% my fault.
[02:07.930 --> 02:08.810] [SPEAKER_00]: Utterly your fault.
[02:08.970 --> 02:15.170] [SPEAKER_00]: And I'm going to let you take the lead on telling the story, Gila, because I was there, but it is primarily your story.
[02:15.170 --> 02:31.250] [SPEAKER_04]: So, the reason that we were not on air last week is because Emmanuel and Kyle are away on assignment, and when Emmanuel and Kyle are away on assignment, their studio is not where the show can happen.
[02:31.250 --> 02:34.910] [SPEAKER_04]: So, the show was set to run from our house.
[02:35.850 --> 02:43.290] [SPEAKER_04]: And we're excited to announce that there does not appear to be a thunderstorm in the offing tonight, as there was the last time we ran the show from here.
[02:43.530 --> 02:45.310] [SPEAKER_04]: So, everybody get excited about that.
[02:46.350 --> 02:49.850] [SPEAKER_04]: So, yeah, we were all set to go ahead and do that from here.
[02:50.530 --> 03:03.950] [SPEAKER_04]: And then Tuesday, Monday night into Tuesday, I was in horrible, horrible, horrible pain, and I couldn't sleep, and I was throwing up, and it was bad.
[03:03.950 --> 03:09.190] [SPEAKER_04]: And we set off for the emergency room on Tuesday afternoon.
[03:09.830 --> 03:15.670] [SPEAKER_04]: I was admitted to the hospital around 1 a.m., I want to say.
[03:16.350 --> 03:25.010] [SPEAKER_04]: And I got up to a room about 9 a.m., and they said, you're having emergency surgery at some point soon.
[03:27.550 --> 03:29.910] [SPEAKER_04]: You know, tell your gallbladder it's been fun.
[03:30.470 --> 03:32.850] [SPEAKER_04]: And that is what happened.
[03:32.850 --> 03:41.570] [SPEAKER_04]: And I wound up having surgery first thing on Thursday morning, and I have been recuperating since.
[03:43.670 --> 03:55.990] [SPEAKER_00]: And obviously, we have to thank our pals at the station, Reggie, management, everyone for being on point and filling in with something when we, at the last moment, couldn't produce a show.
[03:57.570 --> 04:10.230] [SPEAKER_00]: And, yeah, I mean, listeners to this program, regular listeners will know that we are pretty much the backup base when Emmanuel and Kyle can't run things from their pile of technology.
[04:10.230 --> 04:13.250] [SPEAKER_00]: So, yeah, we're doing our best.
[04:13.550 --> 04:18.910] [SPEAKER_00]: And the hospital visit itself, because I was there, I stayed with you.
[04:18.970 --> 04:20.170] [SPEAKER_04]: You're very good, yes.
[04:20.370 --> 04:29.130] [SPEAKER_00]: And the visit itself was very interesting from the standpoint of somebody who talks about the sorts of things we talk about on this program.
[04:30.550 --> 04:52.190] [SPEAKER_04]: I mean, let's just begin with the fact that, obviously, when you're in the hospital, when you're in the emergency department, when you're there and are waiting and they're just throwing things at you to see what they can fix and what will get better, they obviously want to check that they're giving treatments and medication to the right person,
[04:52.190 --> 04:56.610] [SPEAKER_04]: so you keep getting asked your name and your date of birth a lot.
[04:57.790 --> 05:14.970] [SPEAKER_04]: I would like to take this moment to encourage everyone who finds themselves in the emergency room system and has their wits about them enough to be stating their name and date of birth to please use your inside voice when you're stating your PII all over the place.
[05:17.170 --> 05:22.890] [SPEAKER_04]: There was a woman who was in a station across the hallway?
[05:23.430 --> 05:23.590] [SPEAKER_04]: Room?
[05:23.790 --> 05:25.310] [SPEAKER_04]: I'm not even quite sure how to describe it.
[05:25.570 --> 05:28.330] [SPEAKER_04]: The little Warren of stations.
[05:28.790 --> 05:28.930] [SPEAKER_03]: Right.
[05:29.290 --> 05:32.850] [SPEAKER_04]: Who just kept shouting her name and her birth date.
[05:34.550 --> 05:37.190] [SPEAKER_00]: Kind of like a Pokemon, just shouting their own name over and over.
[05:37.190 --> 05:37.310] [SPEAKER_04]: Yeah.
[05:37.710 --> 05:38.610] [SPEAKER_04]: Or like DJ Khaled.
[05:38.850 --> 05:39.090] [SPEAKER_00]: Right.
[05:39.470 --> 05:43.630] [SPEAKER_04]: Linda, if you're listening, I hope your birthday in April is great.
[05:45.350 --> 05:47.630] [SPEAKER_04]: You're going to be...
[05:50.210 --> 05:50.730] [SPEAKER_04]: 67.
[05:53.450 --> 05:55.910] [SPEAKER_04]: And again, just...
[05:55.910 --> 05:56.550] [SPEAKER_04]: Y'all.
[05:56.790 --> 05:57.050] [SPEAKER_00]: Yeah.
[05:57.190 --> 05:59.930] [SPEAKER_04]: If you can keep your wits about you, maybe don't shout.
[06:00.430 --> 06:00.930] [SPEAKER_00]: Okay.
[06:01.090 --> 06:04.730] [SPEAKER_00]: And in addition to that, so there were other folks being attended to.
[06:05.010 --> 06:10.090] [SPEAKER_00]: You got attended to, brought from the emergency room to a proper room and...
[06:10.090 --> 06:11.310] [SPEAKER_00]: And everything.
[06:11.750 --> 06:16.810] [SPEAKER_00]: And as people who have been hospitalized will know, you sometimes get neighbors.
[06:18.950 --> 06:21.890] [SPEAKER_00]: And we had some neighbors in your room.
[06:22.150 --> 06:22.870] [SPEAKER_04]: We did.
[06:23.290 --> 06:27.810] [SPEAKER_00]: And, you know, we were talking to your doctors about everything that needed to be done.
[06:27.810 --> 06:29.310] [SPEAKER_00]: And everything was fairly straightforward.
[06:29.630 --> 06:31.690] [SPEAKER_00]: And, you know, they took, I think, good care of you.
[06:33.070 --> 06:42.190] [SPEAKER_00]: The folks across the room from us were arguing with their doctor of the person who was being seen to.
[06:43.610 --> 06:56.610] [SPEAKER_00]: Because whenever their doctor told them something about what they needed to do to, you know, keep up their treatment, to prepare themselves to do this, to do that, they had a refrain that they kept coming back with and arguing with this doctor.
[06:56.850 --> 06:57.730] [SPEAKER_00]: And, Gila, what was that refrain?
[06:57.730 --> 06:59.530] [SPEAKER_04]: Well, ChatGPT said something else.
[06:59.970 --> 07:03.010] [SPEAKER_04]: I asked ChatGPT and ChatGPT said something else.
[07:03.210 --> 07:11.210] [SPEAKER_04]: So, first of all, a reminder that medical professionals are, in fact, medical professionals with years of training.
[07:11.910 --> 07:16.990] [SPEAKER_04]: And they know what's up, and you don't need to ask ChatGPT everything.
[07:17.130 --> 07:22.190] [SPEAKER_04]: I did hear one of their doctors give maybe the best response I've ever heard.
[07:22.350 --> 07:22.710] [SPEAKER_02]: Mm-hmm.
[07:23.070 --> 07:35.690] [SPEAKER_04]: Which was, ChatGPT scans everything on the Internet and doesn't differentiate between what makes sense and what doesn't make sense.
[07:35.750 --> 07:39.090] [SPEAKER_04]: So, it's important to not listen to them.
[07:39.810 --> 07:40.430] [SPEAKER_00]: Mm-hmm.
[07:41.390 --> 07:46.570] [SPEAKER_04]: And, again, that was the most polite response I could have thought of anyone giving in that moment.
[07:46.790 --> 07:48.010] [SPEAKER_00]: It was very well put.
[07:48.270 --> 07:57.750] [SPEAKER_00]: And, you know, even before the current wave of AI hype, it was a thing for people to consult Dr. Google.
[07:58.030 --> 07:58.390] [SPEAKER_02]: Mm-hmm.
[07:58.610 --> 08:01.270] [SPEAKER_00]: Or, you know, whatever search engine you have.
[08:01.450 --> 08:07.110] [SPEAKER_00]: And just use that to diagnose yourself, to, you know, make your own medical decisions.
[08:07.110 --> 08:18.930] [SPEAKER_00]: And at the end of the day, if you have a trained professional in front of you telling you something is so, and in order to continue your existence on this planet of ours, you need to do certain things.
[08:20.170 --> 08:22.370] [SPEAKER_00]: Maybe don't come back with what ChatGPT said.
[08:22.930 --> 08:23.030] [SPEAKER_04]: Right.
[08:23.170 --> 08:25.390] [SPEAKER_04]: But these are also the people who we heard.
[08:25.530 --> 08:30.750] [SPEAKER_04]: And, again, I would also like to mention that if you're visiting someone in the hospital, your indoor voice is a really good tool.
[08:31.990 --> 08:33.330] [SPEAKER_00]: Boisterous family, they were.
[08:33.570 --> 08:34.210] [SPEAKER_04]: Boisterous family.
[08:35.690 --> 08:46.570] [SPEAKER_04]: Nothing has made me so fearful for the future, as the patient's son very proudly telling his mother, well, I get all my news from ChatGPT and Libertarian podcasts.
[08:51.670 --> 08:58.430] [SPEAKER_04]: And, you know, it was a positive way to keep from, to keep distracted.
[08:59.650 --> 09:03.750] [SPEAKER_04]: I'm excited to not think about whatever else was going on.
[09:04.530 --> 09:05.830] [SPEAKER_04]: I appreciated it.
[09:06.610 --> 09:08.970] [SPEAKER_04]: But, yeah, indoor voices are good.
[09:10.050 --> 09:11.230] [SPEAKER_00]: Indoor voices are good.
[09:11.510 --> 09:15.830] [SPEAKER_00]: And the information given you by trained experts in their field trumps.
[09:16.390 --> 09:29.850] [SPEAKER_00]: Somebody recently, and I saw this, I think, from multiple posters over the past couple of weeks, posted something to the effect of, whenever somebody's statement to me begins, I ask ChatGPT, I know I'm about to hear some dumbass crap.
[09:31.150 --> 09:31.570] [SPEAKER_04]: Yeah.
[09:32.890 --> 09:37.590] [SPEAKER_00]: But, you know, when life or death is involved, maybe use a little more discretion.
[09:38.070 --> 09:39.250] [SPEAKER_00]: Alex, how are you doing out there?
[09:39.390 --> 09:41.870] [SPEAKER_00]: You seem to be located locally this week.
[09:42.870 --> 09:43.570] [SPEAKER_01]: I am.
[09:43.690 --> 09:44.590] [SPEAKER_01]: I'm in New York.
[09:44.590 --> 09:54.710] [SPEAKER_01]: I'm just being bedazzled by the story of Gila and this hospital experience over here and remaining somewhat quiet.
[09:54.930 --> 09:57.570] [SPEAKER_01]: But I tend to agree with you, though, Rob.
[09:58.090 --> 10:09.930] [SPEAKER_01]: And I have found, too, that in my experience being a lawyer and negotiating with people, you know, ChatGPT is just used by counterparties all the time.
[10:10.070 --> 10:11.030] [SPEAKER_01]: It might be ChatGPT.
[10:11.030 --> 10:12.790] [SPEAKER_01]: You might be, you know, some other LLM.
[10:13.010 --> 10:20.810] [SPEAKER_01]: But I often feel like I'm always just negotiating with a large language model or some kind of generative AI at this point.
[10:21.030 --> 10:24.670] [SPEAKER_01]: And it makes people think that they know better than they do.
[10:24.810 --> 10:29.250] [SPEAKER_01]: So it is a little bit terrifying to see this being used in a hospital environment.
[10:29.750 --> 10:39.090] [SPEAKER_01]: Thankfully, and knock on wood here, I haven't been in a hospital since the advent of, you know, massive large language models have come into play.
[10:39.090 --> 10:41.090] [SPEAKER_01]: But crazy.
[10:41.810 --> 10:42.570] [SPEAKER_01]: What an experience.
[10:42.570 --> 10:46.730] [SPEAKER_01]: We're really glad that you're back and seemingly okay, Gila.
[10:46.890 --> 10:47.110] [SPEAKER_01]: Thank you.
[10:47.350 --> 10:49.030] [SPEAKER_01]: How is everything now?
[10:49.950 --> 10:50.850] [SPEAKER_04]: Things are good.
[10:50.990 --> 10:56.630] [SPEAKER_04]: I mean, the fascinating thing about the gallbladder is that it is useless.
[10:57.950 --> 11:07.310] [SPEAKER_04]: And when, you know, when you're healing from so many surgeries, it's about your body learning to do something differently or work without something that used to be there.
[11:08.010 --> 11:10.870] [SPEAKER_04]: And in this case, it's just closing up a hole.
[11:11.970 --> 11:12.790] [SPEAKER_04]: It doesn't...
[11:12.790 --> 11:19.770] [SPEAKER_04]: My body doesn't need to learn how to do anything new because the gallbladder is still so vestigial and dumb.
[11:24.650 --> 11:29.270] [SPEAKER_04]: I'm getting to the point where I'm bored, which is bad, because then I push too hard and then I'm exhausted.
[11:30.210 --> 11:31.750] [SPEAKER_04]: But this is amazing to me.
[11:32.070 --> 11:35.110] [SPEAKER_04]: This whole process was a week ago.
[11:35.230 --> 11:35.570] [SPEAKER_04]: Not even...
[11:35.570 --> 11:37.050] [SPEAKER_04]: Like, a week ago, I hadn't even had surgery yet.
[11:38.390 --> 11:43.250] [SPEAKER_04]: And that is just bewildering to me.
[11:45.410 --> 11:51.450] [SPEAKER_04]: You know, I am lucky in so many ways that things turned out the way that they did.
[11:52.470 --> 12:01.470] [SPEAKER_04]: Not least that there is an emergency room literally two blocks from our apartment that we were able to get to on foot and pretty quickly.
[12:03.450 --> 12:03.910] [SPEAKER_00]: Yeah.
[12:05.990 --> 12:14.070] [SPEAKER_00]: And talking of this whole experience, so, you know, you're here, you're convalescing, not at all through the efforts of large language models.
[12:16.270 --> 12:18.310] [SPEAKER_00]: But, yeah, you're doing well.
[12:18.470 --> 12:23.150] [SPEAKER_00]: I can confirm for the folks out there, as someone who's in the room with you, you're doing extraordinarily well.
[12:24.030 --> 12:25.790] [SPEAKER_00]: And just keep taking care of yourself.
[12:26.150 --> 12:32.150] [SPEAKER_00]: And this whole ChatGPT thing does connect with a bunch of stuff that's been in the news the past couple of weeks.
[12:33.370 --> 12:35.170] [SPEAKER_00]: I wanted to talk about this one story.
[12:35.170 --> 12:37.390] [SPEAKER_00]: We didn't get to it last week because we weren't around.
[12:38.450 --> 12:39.950] [SPEAKER_00]: This is from the 5th of December.
[12:40.650 --> 12:52.790] [SPEAKER_00]: A BBC news story that trains were cancelled because somebody had started circulating generative AI images of the bridge collapsing that this train was going to go over.
[12:53.770 --> 12:54.830] [SPEAKER_00]: It was...
[12:56.910 --> 12:58.450] [SPEAKER_00]: There was a tremor.
[12:58.630 --> 13:02.550] [SPEAKER_00]: And there was a tremor which struck on Wednesday night.
[13:02.690 --> 13:05.070] [SPEAKER_00]: I guess Wednesday night of the week of the 5th.
[13:05.150 --> 13:06.350] [SPEAKER_00]: That would have been the 3rd.
[13:06.870 --> 13:09.210] [SPEAKER_00]: Felt across Lancashire in the Southern Lake District.
[13:10.410 --> 13:14.090] [SPEAKER_00]: And this image started being circulated of the bridge, collapsed.
[13:14.350 --> 13:16.370] [SPEAKER_00]: And of course, in reality, it was fine.
[13:16.670 --> 13:25.290] [SPEAKER_00]: But the realism of this image was enough for them to start canceling trains and, you know, all that sort of thing.
[13:25.490 --> 13:33.550] [SPEAKER_00]: And, yeah, I found this story a little interesting just as an example of the power of this sort of thing now.
[13:33.610 --> 13:40.170] [SPEAKER_00]: We are officially at a point where generative AI video can look convincing, at least to an untrained eye.
[13:40.810 --> 13:46.970] [SPEAKER_04]: But the idea that the BBC then ran the video through AI to determine where the AI was.
[13:47.470 --> 13:47.910] [SPEAKER_00]: Yes.
[13:48.090 --> 13:53.010] [SPEAKER_04]: That may be my favorite piece of this little story.
[13:53.210 --> 13:53.330] [SPEAKER_00]: Yeah.
[13:53.450 --> 13:59.370] [SPEAKER_00]: The story on BBC.com does say, a BBC journalist ran the image through an AI chatbot which identified key spots that may have been manipulated.
[13:59.650 --> 14:02.070] [SPEAKER_00]: This is, of course, a stupid thing to do because it means nothing.
[14:02.630 --> 14:03.110] [SPEAKER_04]: Right.
[14:04.330 --> 14:11.930] [SPEAKER_04]: But, again, I think the tendency to want to hand so much of automation over to AI...
[14:11.930 --> 14:23.610] [SPEAKER_04]: I mean, there was another story that came up about that weapons recognition system that has caught two kids with chip bags.
[14:23.970 --> 14:27.730] [SPEAKER_00]: Yeah, I think one kid had a chip bag and another kid had a clarinet.
[14:28.690 --> 14:30.390] [SPEAKER_04]: Yeah, the most recent one was a clarinet.
[14:30.390 --> 14:34.630] [SPEAKER_04]: And apparently the child was holding a clarinet like a gun.
[14:36.690 --> 14:42.530] [SPEAKER_04]: So, the official word from school leadership is, well, this is working exactly the way it's supposed to.
[14:42.870 --> 14:47.790] [SPEAKER_04]: And you shouldn't hold your clarinet like a gun because the system is going to...
[14:49.930 --> 14:51.170] [SPEAKER_04]: I don't know, man.
[14:51.310 --> 14:54.910] [SPEAKER_04]: When you were a little kid, did you ever not hold anything like a gun?
[14:55.350 --> 14:55.870] [SPEAKER_00]: Oh, yeah.
[14:56.030 --> 14:56.750] [SPEAKER_04]: Everything was a gun.
[14:56.970 --> 14:57.570] [SPEAKER_00]: Everything was...
[14:57.570 --> 14:59.190] [SPEAKER_00]: Absolutely everything on the playground was a gun.
[15:00.390 --> 15:04.470] [SPEAKER_00]: Pew-pew lasers or, you know, boom-boom incendiary weapons.
[15:05.330 --> 15:06.510] [SPEAKER_00]: But, yeah.
[15:06.790 --> 15:10.030] [SPEAKER_00]: And clarinets are commonly found around schools.
[15:10.030 --> 15:12.550] [SPEAKER_00]: I think just as common as chip bags, maybe.
[15:14.150 --> 15:14.670] [SPEAKER_00]: Yeah.
[15:14.870 --> 15:17.990] [SPEAKER_00]: So, maybe the technology needs to continue to be refined.
[15:18.270 --> 15:33.290] [SPEAKER_00]: And, of course, this brings into sharp relief the danger of trusting these things too much, especially if you're an understaffed authority figure who just wants to somehow throw a switch and believe things are taken care of that aren't.
[15:34.130 --> 15:40.590] [SPEAKER_04]: Do we think the person who created the image or who prompted the image, I guess, more to the point, is proud?
[15:43.290 --> 15:43.730] [SPEAKER_00]: Maybe.
[15:44.090 --> 15:45.570] [SPEAKER_00]: There are always going to be trolls, aren't there?
[15:46.150 --> 15:46.730] [SPEAKER_00]: There are always...
[15:46.730 --> 15:49.010] [SPEAKER_00]: And trolls and bridges are a natural match.
[15:49.130 --> 15:49.770] [SPEAKER_00]: Oh, we did that.
[15:49.770 --> 15:50.710] [SPEAKER_00]: Oh, yes.
[15:51.810 --> 15:52.730] [SPEAKER_00]: Alex, yes.
[15:53.570 --> 16:11.710] [SPEAKER_01]: Well, I think this brings up an interesting topic because we certainly have gotten past the point now where people have started to distrust quite literally everything that they see on social media because you're not sure whether or not something is authentic or it's AI-generated.
[16:12.010 --> 16:28.910] [SPEAKER_01]: Any reel that looks unreal or is somewhat unbelievable on Instagram or on Facebook, if you look at the comments on this, people claiming it's AI, other people claiming it's real, there's genuine confusion out there about what is authentic and what is inauthentic.
[16:29.290 --> 16:46.930] [SPEAKER_01]: And it brings to mind the need to have some kind of clarity about this, either from a platform, but it becomes difficult to detect, I think, real artifacts of whether or not something is going to be an AI-generated image or a legitimate image.
[16:47.430 --> 16:52.910] [SPEAKER_01]: It brings to mind the work of somebody by the name of Hani Farid.
[16:53.150 --> 16:58.430] [SPEAKER_01]: And I used to work with Hani at Dartmouth a long time ago.
[16:58.450 --> 17:06.990] [SPEAKER_01]: My God, I can't believe it was about 20 years ago now at a federally funded think tank called the Institute for Security Technology Studies.
[17:07.090 --> 17:10.250] [SPEAKER_01]: I think it's called the Institute for Security Technology and Society now.
[17:10.250 --> 17:23.650] [SPEAKER_01]: But Hani Farid was a computer science professor at Dartmouth, and he got into this field of detecting altered images by trying to detect images that had embedded steganographic encryption.
[17:24.450 --> 17:50.690] [SPEAKER_01]: So if you remember, shortly after 9-11, there was a lot of panic about the potential for terrorists to post a picture of a couch or something for sale on Facebook and having that couch image a hidden message that related to where some plot should begin or a message like the skeletons dance at midnight on some crappy couch that's for sale on Craigslist or something like that,
[17:50.750 --> 17:50.890] [SPEAKER_01]: right?
[17:51.430 --> 17:58.870] [SPEAKER_01]: So Hani Farid started to figure out ways to detect whether or not certain pixels were altered at very low levels.
[17:59.030 --> 18:07.450] [SPEAKER_01]: And he's now left Dartmouth and teaches at UC Berkeley, but he's a real pioneer on digital image forensics.
[18:07.770 --> 18:28.350] [SPEAKER_01]: And what he's done is his work, to my mind, is all about really low-level pixel analysis and looking at physical and geometric consistency and to determine whether or not an image has been AI generated by looking at, you know, what real geometry should be and what it actually is.
[18:28.750 --> 18:32.090] [SPEAKER_01]: And he's built systems that do this really, really well.
[18:32.090 --> 18:46.290] [SPEAKER_01]: And it seems to me like this type of thing needs to be implemented on all of the platforms out there to determine whether or not something has the hallmarks or the indications and indicia of it being AI generated or not.
[18:46.390 --> 18:52.110] [SPEAKER_01]: Because we are way past this point of believability when it comes to images.
[18:52.310 --> 18:53.650] [SPEAKER_01]: And it's becoming a real problem.
[18:53.790 --> 18:58.230] [SPEAKER_01]: And if we're to trust anything in the future, we need to know what's authentic and what's inauthentic.
[18:59.010 --> 19:00.790] [SPEAKER_04]: And actually, can I roll back for a second?
[19:01.130 --> 19:04.710] [SPEAKER_04]: Because talking about images, I do have a bit more detail.
[19:05.230 --> 19:05.330] [SPEAKER_00]: Yeah.
[19:05.910 --> 19:09.590] [SPEAKER_04]: The reason I had to have my gallbladder out was because I had gallstones.
[19:09.990 --> 19:17.830] [SPEAKER_04]: And right before I went into surgery, I said to the surgeon very politely, would it be possible to get a picture if you remember?
[19:18.430 --> 19:19.230] [SPEAKER_04]: I'm curious.
[19:19.230 --> 19:20.670] [SPEAKER_04]: I'd like to see what's going on.
[19:21.250 --> 19:23.370] [SPEAKER_04]: And she said, I'll do it if I remember.
[19:23.690 --> 19:24.910] [SPEAKER_04]: And I said, okay, cool.
[19:26.010 --> 19:27.030] [SPEAKER_04]: And then I had surgery.
[19:27.310 --> 19:33.370] [SPEAKER_04]: And later I was in my room and the surgeon came by to check in and she said, I remembered to take a picture.
[19:33.970 --> 19:35.670] [SPEAKER_04]: That was really something.
[19:37.230 --> 19:38.670] [SPEAKER_04]: And I said, what do you mean?
[19:38.750 --> 19:42.250] [SPEAKER_04]: And she said, you had one gigantic gallstone.
[19:44.410 --> 19:50.630] [SPEAKER_04]: We, she had put a pen with like a little ruler on it next to said gallstone.
[19:51.470 --> 19:52.730] [SPEAKER_04]: It was, how big was it?
[19:53.090 --> 20:01.210] [SPEAKER_00]: For the photo, yeah, it was four centimeters long, about 2.7 centimeters wide, egg-shaped, as these things are.
[20:01.450 --> 20:01.890] [SPEAKER_04]: Enormous.
[20:02.330 --> 20:08.350] [SPEAKER_04]: And she airdropped this photo to me from her phone to my phone.
[20:08.990 --> 20:12.590] [SPEAKER_04]: And for a moment, I was like, this can't be real.
[20:13.710 --> 20:17.110] [SPEAKER_04]: Not that I had any sense of what it was supposed to look like.
[20:17.190 --> 20:20.610] [SPEAKER_04]: Not that I had any sense of how many gallstones I had or where they were supposed to be.
[20:21.810 --> 20:31.930] [SPEAKER_04]: But I have spent a not insignificant amount of time over the last week trying to kind of reverse image search and make sure that this is really what was there.
[20:32.230 --> 20:34.550] [SPEAKER_04]: It is terrifyingly enormous, I must say.
[20:34.730 --> 20:37.750] [SPEAKER_00]: It's an achievement on the surgeon's part and on your part.
[20:38.050 --> 20:39.470] [SPEAKER_00]: I'm very proud of myself for accomplishing that.
[20:39.630 --> 20:39.910] [SPEAKER_00]: Well crafted.
[20:41.490 --> 20:42.830] [SPEAKER_04]: years of work, apparently.
[20:43.110 --> 20:46.430] [SPEAKER_00]: I always respect your creative forces and outputs.
[20:47.970 --> 20:51.170] [SPEAKER_00]: So I wanted to talk about some more related stuff.
[20:51.750 --> 21:06.490] [SPEAKER_00]: We were talking about AI and also, you know, the general media environment because I think in all our lifetimes, those of us on this program, probably everyone listening to it, photos have been something that, you know, if someone hands you a paper photo and says,
[21:06.590 --> 21:12.270] [SPEAKER_00]: this is proof of something, it's been common knowledge that, well, you can fake photos, you can alter photos.
[21:12.650 --> 21:17.970] [SPEAKER_00]: And this used to be something very complex and very difficult to do unless you had certain talents and equipment and stuff.
[21:18.210 --> 21:20.450] [SPEAKER_00]: But over the years, it got easier and easier and easier.
[21:20.610 --> 21:27.650] [SPEAKER_00]: So if somebody handed you a photo, maybe it's Photoshop, maybe it's real, but you would never assume that it was the entire truth.
[21:28.330 --> 21:37.430] [SPEAKER_00]: But video was more dependable because video contains many images and it's so much more difficult to edit video than it is to edit an image.
[21:37.530 --> 21:45.290] [SPEAKER_00]: You know, anybody can Photoshop something, but video was more of a solid thing and we are now past that point.
[21:45.950 --> 21:47.610] [SPEAKER_00]: Video is not proof of anything.
[21:47.930 --> 21:49.550] [SPEAKER_00]: Video is easily fakeable.
[21:50.430 --> 22:02.710] [SPEAKER_00]: So, yeah, we all have to be sort of on our guard with stuff that we didn't even five, six years ago on levels because anybody can use these tools and anybody can generate this stuff.
[22:03.030 --> 22:05.790] [SPEAKER_00]: So, yeah, be wary out there.
[22:05.950 --> 22:06.910] [SPEAKER_00]: Consider your sources.
[22:07.530 --> 22:09.030] [SPEAKER_00]: And use your judgment.
[22:10.530 --> 22:13.690] [SPEAKER_00]: Another story going around this week involving AI.
[22:13.870 --> 22:18.590] [SPEAKER_00]: Microsoft Copilot is an AI assistant as these things go.
[22:18.770 --> 22:20.370] [SPEAKER_04]: I uninstalled it for my laptop this afternoon.
[22:20.370 --> 22:21.330] [SPEAKER_00]: Nicely done.
[22:22.350 --> 22:27.070] [SPEAKER_00]: And you could do that because your laptop is something that you have a reasonable amount of control over.
[22:27.630 --> 22:43.870] [SPEAKER_00]: If you were the owner of a fancy LG Smart television, you may find yourself with that problem but unable to address it because LG Smart TV owners are reporting that a recent webOS software update has added Microsoft Copilot to their TVs with no apparent way to remove it.
[22:43.870 --> 22:45.730] [SPEAKER_00]: this story from Tom's Hardware.
[22:46.710 --> 22:51.750] [SPEAKER_00]: Reports first surfaced over the weekend on Reddit where a post showed a copilot tile pinned to an LG TV home screen.
[22:53.090 --> 22:56.670] [SPEAKER_00]: And apparently this is now installed on people's televisions.
[22:56.690 --> 22:57.930] [SPEAKER_00]: It cannot be removed.
[22:58.450 --> 22:59.710] [SPEAKER_00]: It can't be uninstalled.
[22:59.890 --> 23:02.210] [SPEAKER_00]: Not through any way that people have found so far.
[23:02.210 --> 23:05.450] [SPEAKER_00]: and a lot of people are angry about it.
[23:07.810 --> 23:22.090] [SPEAKER_00]: There are so many stories going around of smart devices like being subverted, being used to spy on people and here is something that gets stuck on your very expensive device without your permission.
[23:22.670 --> 23:37.390] [SPEAKER_04]: And apparently the only way you can disable Copilot right now is to disconnect your TV from the Internet which given the state of the way that so much of entertainment works right now means you can't do anything else either.
[23:38.390 --> 23:40.110] [SPEAKER_04]: So then you turn on the fireplace and go.
[23:41.390 --> 23:42.870] [SPEAKER_00]: Or just set the TV on fire.
[23:43.010 --> 23:44.830] [SPEAKER_00]: Maybe that'll entertain as well.
[23:45.270 --> 23:56.590] [SPEAKER_00]: But yeah, if you're running apps on your television, those need Internet access for your streaming services or whatever else, your digital fish tank, whatever you got going on.
[23:57.230 --> 24:02.530] [SPEAKER_00]: And the whole smart TV scene has always been kind of a dangerous place full of dodgy apps and things.
[24:02.730 --> 24:06.510] [SPEAKER_00]: But this is above and beyond.
[24:06.970 --> 24:13.330] [SPEAKER_00]: And I don't know, Alex, you're a TV watcher yourself at times.
[24:13.830 --> 24:18.090] [SPEAKER_00]: How would you feel if you had something like this added to your television and you couldn't get rid of it?
[24:18.770 --> 24:19.890] [SPEAKER_01]: Oh, I mean, wouldn't it be great?
[24:20.090 --> 24:23.290] [SPEAKER_01]: Because doesn't AI just make our lives so much easier?
[24:23.510 --> 24:31.270] [SPEAKER_01]: I mean, how could you not want an AI system tracking all of your watching habits and maybe offering you suggestions?
[24:31.430 --> 24:35.590] [SPEAKER_01]: I mean, I know how much you love friends over there, Rob.
[24:35.830 --> 24:43.430] [SPEAKER_01]: And so, you know, it could also offer you suggestions for, you know, other related episodes of Friends or I'm joking, of course, Rob.
[24:43.570 --> 24:46.170] [SPEAKER_01]: It's not an enthusiast over there.
[24:47.130 --> 24:49.030] [SPEAKER_01]: And I'm being sarcastic.
[24:49.030 --> 24:50.430] [SPEAKER_01]: I think the whole thing is deplorable.
[24:50.650 --> 24:57.750] [SPEAKER_01]: I think it raises a lot of issues about how and why this is being done.
[24:58.210 --> 25:11.650] [SPEAKER_01]: I mean, putting these tiles into somebody else's television, some product that you own, I don't understand why companies continue to do stupid things like this because it's just going to make users angry.
[25:11.870 --> 25:12.910] [SPEAKER_01]: It's going to make people angry.
[25:13.090 --> 25:22.770] [SPEAKER_01]: Nobody, I don't know, maybe some people will be thankful for it in much the same way that I was tongue-in-cheek being thankful for the installation of Copilot on the television.
[25:22.950 --> 25:24.130] [SPEAKER_01]: But I doubt it.
[25:25.150 --> 25:28.210] [SPEAKER_01]: And I think for a good portion of users, they are going to be angry.
[25:28.230 --> 25:39.070] [SPEAKER_01]: And for a good portion of users, it raises the issue of would they have consented to this at the outset of the bargain here of when they bought the television.
[25:39.710 --> 25:44.550] [SPEAKER_01]: Because if you're not consenting to the collection of your data, they shouldn't be doing it.
[25:44.910 --> 25:47.650] [SPEAKER_01]: So how are they doing this comes to mind.
[25:47.970 --> 26:00.990] [SPEAKER_01]: They very likely had to change the terms of service or the terms of use so that in order for you to continue using the television, you probably had to accept these new terms of service and then Microsoft Copilot presumably was installed.
[26:01.190 --> 26:05.050] [SPEAKER_01]: That's my guess as to what happened here with these LG television sets.
[26:05.050 --> 26:20.790] [SPEAKER_01]: So those contracts where the terms of service are updated or you can't and you have to accept those changes unless you decide to stop using that television at all, those are really termed contracts of adhesion.
[26:20.990 --> 26:27.090] [SPEAKER_01]: You don't necessarily have any kind of bargaining power to change those particular terms with LG.
[26:27.390 --> 26:31.230] [SPEAKER_01]: You can't go and talk to their lawyers and say, hey, I would like to remove this particular clause.
[26:31.230 --> 26:33.650] [SPEAKER_01]: and so it's a contract of adhesion.
[26:33.910 --> 26:54.910] [SPEAKER_01]: So with those terms of service updated through a contract of adhesion, I think that the original bargain between the purchaser of the television and LG self has been materially changed and it's been changed midstream while somebody is using this particular television.
[26:55.150 --> 27:09.030] [SPEAKER_01]: So in some respects, I could see this type of addition of an AI system potentially opening up LG or other companies that would do something like this to liability because it is a consent issue.
[27:09.310 --> 27:24.630] [SPEAKER_01]: And I'm sure that there are some crafty plaintiff's lawyers very likely in California because that's where all of these lawsuits are filed that are already putting the final touches on a class action complaint for some kind of privacy breach under the CCPA,
[27:24.870 --> 27:33.690] [SPEAKER_01]: the California Consumer Privacy Act or other state laws that would relate to data collection and consent practices for this particular type of situation.
[27:34.050 --> 27:35.590] [SPEAKER_01]: So I think it's a bit hairy.
[27:35.890 --> 27:39.790] [SPEAKER_01]: It seems like it's just an annoying small thing, but it could have major implications.
[27:41.070 --> 27:57.370] [SPEAKER_00]: Yeah, and funny you mention lawsuits regarding this sort of thing because there's another story, this one published yesterday on TechRadar, that Texas is considering themselves messed with by LG and also by Samsung and Hisense and TCL and other manufacturers of smart TVs.
[27:57.990 --> 28:10.890] [SPEAKER_00]: Texas Attorney General Ken Paxton filed a suit on Monday against those companies, claiming they have been unlawfully collecting personal data through something called ACR, which stands for Automated Content Recognition.
[28:11.710 --> 28:14.390] [SPEAKER_00]: And, you know, that's exactly what it sounds like.
[28:14.490 --> 28:25.710] [SPEAKER_00]: Whatever you're watching on your TV, whether you're watching it from an app on the TV, whether you're watching it from something connected to the TV, from your home video device, from, you know, pirated files, from whatever, what have you.
[28:25.890 --> 28:30.870] [SPEAKER_00]: It is sampling what's being shown on your television and phoning home with that information.
[28:33.450 --> 28:36.610] [SPEAKER_00]: The now more than decade-old technology has a long history.
[28:37.130 --> 28:42.910] [SPEAKER_00]: And another company, Samba TV, patented its own ACR tech in 2011.
[28:43.990 --> 28:47.550] [SPEAKER_00]: Most of the early uses revolved around sound and music recognition.
[28:47.650 --> 28:55.950] [SPEAKER_00]: If you use the app Shazam to tell what music is playing on, you know, you know, within your field of hearing, it's using automated content recognition.
[28:56.010 --> 28:59.570] [SPEAKER_00]: But this is also being done on smart televisions.
[29:00.030 --> 29:10.370] [SPEAKER_00]: And not everyone wants what they're watching to be sent to who knows who, whether it's the TV manufacturer, whether it's some, you know, companies that pay the TV manufacturer for that information.
[29:12.730 --> 29:26.470] [SPEAKER_00]: And, you know, Texas is the Attorney General of Texas, who is somebody named Ken Paxton, who has his own long list of controversies, if you want to check out his Wikipedia article, more than worth going into here.
[29:26.610 --> 29:31.650] [SPEAKER_00]: But even Texans who don't like him are considering this maybe sort of a stop clock moment.
[29:31.930 --> 29:33.070] [SPEAKER_04]: I was thinking exactly stop clocks.
[29:33.250 --> 29:33.530] [SPEAKER_04]: Absolutely.
[29:33.930 --> 29:34.270] [SPEAKER_00]: Yes.
[29:34.430 --> 29:36.070] [SPEAKER_00]: As in a stop clock is right twice a day.
[29:36.550 --> 29:38.370] [SPEAKER_00]: Um, so yes.
[29:38.370 --> 29:39.330] [SPEAKER_04]: I have a question though.
[29:39.790 --> 29:39.950] [SPEAKER_00]: Sure.
[29:40.090 --> 29:40.710] [SPEAKER_04]: Side note backing up.
[29:40.830 --> 29:40.970] [SPEAKER_00]: Shoot.
[29:41.150 --> 29:47.610] [SPEAKER_04]: Now that everything is digital, and there's so few analog clocks in the world, do people understand the phrase stop clocks are right twice a day?
[29:50.090 --> 29:51.450] [SPEAKER_00]: This is an excellent question.
[29:52.330 --> 29:52.950] [SPEAKER_04]: Thank you.
[29:53.510 --> 29:54.790] [SPEAKER_00]: It is one of those things.
[29:54.930 --> 29:58.590] [SPEAKER_00]: Are kids in school being taught to read an analog clock face?
[29:59.330 --> 30:02.050] [SPEAKER_00]: I know cursive writing is falling out of favor.
[30:02.730 --> 30:03.770] [SPEAKER_00]: Things like that.
[30:04.150 --> 30:05.830] [SPEAKER_00]: And, you know, who knows?
[30:06.050 --> 30:07.450] [SPEAKER_00]: I don't have kids.
[30:07.770 --> 30:09.010] [SPEAKER_00]: Alex, you've got kids.
[30:09.150 --> 30:14.710] [SPEAKER_00]: Have you had moments where you're, like, maybe they're not teaching something that used to be a fundamental like this?
[30:16.110 --> 30:18.870] [SPEAKER_01]: I think script, cursive.
[30:19.070 --> 30:21.970] [SPEAKER_01]: I haven't seen my kids write in script.
[30:22.310 --> 30:22.950] [SPEAKER_01]: That's it.
[30:23.070 --> 30:25.810] [SPEAKER_01]: I mean, they certainly don't know the Pledge of Allegiance as well.
[30:25.990 --> 30:33.770] [SPEAKER_01]: That's something that seems to have come to the wayside, at least in New York City public schools, and I don't imagine it coming back anytime soon.
[30:34.910 --> 30:38.270] [SPEAKER_01]: But, yeah, I don't know.
[30:38.390 --> 30:46.070] [SPEAKER_01]: I would think that they would, though, because I think in terms of telling time, though, they did have to learn how to tell time on analog clocks.
[30:46.270 --> 30:52.170] [SPEAKER_01]: That seems to me like a life skill that would be very fruitful to have throughout.
[30:52.450 --> 31:00.350] [SPEAKER_01]: Otherwise, you're kind of looking at something or seeing this alien circular structure and have no idea what the hell it is.
[31:00.350 --> 31:10.230] [SPEAKER_01]: So I do think they learn how to tell time, and that the stop clock is right twice a day phrase should make sense to anybody over the age of six.
[31:11.110 --> 31:11.810] [SPEAKER_00]: I'm glad.
[31:11.990 --> 31:30.890] [SPEAKER_00]: It does remind me of when we used to go to maker fairs, which are much missed, but the maker fair out in Corona Park where we would set up a booth with our pals from 2600, and among the things we had for display was a rotary phone that people could come up and use.
[31:31.210 --> 31:39.450] [SPEAKER_00]: And it was such a fun thing to watch kids attending the event come up, and we would challenge them, do you know how to dial a number on this phone?
[31:40.910 --> 31:42.130] [SPEAKER_00]: And sometimes they knew.
[31:42.470 --> 31:45.090] [SPEAKER_00]: Sometimes they had seen it in old films or shows or whatever.
[31:45.510 --> 31:49.670] [SPEAKER_00]: Sometimes they would pick it up and not know what to do with the big round thing with the holes.
[31:50.590 --> 31:57.390] [SPEAKER_00]: More than one kid pressed on the circle in the middle of a rotary dial, expecting it to be like a button or a touch screen of some sort.
[31:59.170 --> 32:07.090] [SPEAKER_00]: And in addition to the fun of seeing if the kids could do it was watching their parents realize, oh my God, my kids don't know how to do this.
[32:07.210 --> 32:08.190] [SPEAKER_00]: I'm so old.
[32:08.630 --> 32:21.910] [SPEAKER_04]: So one thing I've noticed, because sometimes you look to see what exists in the world, and there are a lot of rotary look phones that are available for purchase where it's buttons and they're in that setup, but you can't actually move the plate.
[32:22.050 --> 32:23.110] [SPEAKER_00]: Yeah, they're in that circle.
[32:23.210 --> 32:23.830] [SPEAKER_04]: Which makes me sad.
[32:23.830 --> 32:31.430] [SPEAKER_00]: And then there are spots on a rotary dial for the hash mark and asterisk, which just seem wrong somehow.
[32:34.460 --> 32:36.540] [SPEAKER_00]: But yeah, so keep an eye on your TVs.
[32:36.800 --> 32:52.240] [SPEAKER_00]: For the moment, it seems on an LG TV, if you have to dig for what they call the live plus setting, that's their automated content recognition, and you're able to disable it, but it has been known to be on by default, and you will have to dig for this setting.
[32:52.420 --> 32:53.880] [SPEAKER_00]: It is a dark pattern.
[32:53.880 --> 32:55.460] [SPEAKER_00]: It is not made easy for you.
[32:56.400 --> 33:01.440] [SPEAKER_00]: And as we're seeing with the co-pilot thing, that setting is there until they decide to stop giving you the choice.
[33:02.780 --> 33:17.140] [SPEAKER_00]: One question that comes up a lot with stuff like this is, well, why doesn't somebody just make a dumb TV that works like TVs used to, where you just hook a, you know, you hook your stuff up to it, your cable, your video, whatever, and it'll just show you what you send to it.
[33:18.740 --> 33:21.800] [SPEAKER_00]: Why doesn't some company do that since it doesn't seem like you can buy one?
[33:23.880 --> 33:26.480] [SPEAKER_00]: Solid question, but what's going on is much more insidious.
[33:26.540 --> 33:27.620] [SPEAKER_00]: Dumb TVs do exist.
[33:27.800 --> 33:31.480] [SPEAKER_00]: They're known as commercial displays, usually, or commercial monitors, things of that sort.
[33:31.980 --> 33:37.940] [SPEAKER_00]: And you can find them being sold by the same companies that make smart TVs, but they are much more expensive.
[33:38.280 --> 33:48.440] [SPEAKER_00]: A commercial display with a comparable resolution size, you know, hardware feature set, everything to a smart TV will cost more than the smart TV.
[33:49.420 --> 34:05.160] [SPEAKER_00]: And the reason for that is that the companies who do things like throw Microsoft Copilot on there, or who throw their preloaded apps on there to entice you to subscribe to their streaming services, they all pay the manufacturer for the opportunity to get that stuff in front of your eyes.
[34:05.560 --> 34:12.620] [SPEAKER_00]: That payment goes toward lowering the price of the TV, at least some of it, so you're more likely to grab it over a competing product on the shelf.
[34:12.760 --> 34:25.700] [SPEAKER_00]: So the higher price of the commercial display is more correctly thought of as the actual price of the appliance, whereas the smart TV, which you're thinking is so much cheaper and that's, you know, that's so much better.
[34:26.140 --> 34:31.540] [SPEAKER_00]: It's, you know, that difference in price is what the companies are paying for your eyeballs, basically.
[34:33.500 --> 34:35.060] [SPEAKER_04]: What's the word I'm looking for?
[34:37.580 --> 34:38.800] [SPEAKER_00]: I'm going to need some hints.
[34:40.280 --> 34:49.160] [SPEAKER_04]: When a company pays money to help you pay less, there's a word for that, and I can't remember the word right now.
[34:49.320 --> 34:50.400] [SPEAKER_00]: Can you say the word on the radio?
[34:50.720 --> 34:51.060] [SPEAKER_04]: Yeah, I can.
[34:51.220 --> 34:52.940] [SPEAKER_04]: It's definitely a word you can say on the radio.
[34:54.600 --> 34:55.400] [SPEAKER_04]: It's just...
[34:57.160 --> 34:57.860] [SPEAKER_04]: Oh, my God.
[34:58.760 --> 35:00.220] [SPEAKER_00]: Well, it'll come to you.
[35:00.360 --> 35:01.160] [SPEAKER_00]: And Alex, what's up?
[35:02.080 --> 35:10.900] [SPEAKER_01]: Yeah, I mean, so basically you're saying then, Rob, that the sale of the consumer data would contribute to the lower price point.
[35:10.940 --> 35:12.000] [SPEAKER_01]: Is that correct?
[35:12.920 --> 35:18.580] [SPEAKER_00]: Yeah, in a roundabout way, because the manufacturers of the televisions get paid to put that stuff in front of you.
[35:18.760 --> 35:24.260] [SPEAKER_00]: So they're going to put a little bit of that toward lowering the price of the television, probably also still keeping some of it for themselves.
[35:24.260 --> 35:30.220] [SPEAKER_00]: But that's how money is flowing in this sort of equation over...
[35:31.060 --> 35:48.180] [SPEAKER_01]: So it's just another form, essentially, of the surveillance, capitalism, economic principles at play here, where the data that we're generating is valuable, it's being sold, it's being packaged, it's being bundled and quote-unquote shared with third-party affiliates,
[35:48.300 --> 35:48.500] [SPEAKER_01]: etc.
[35:49.300 --> 35:59.440] [SPEAKER_01]: And what's interesting is I think that that all is premised under an assumption that that could be made erroneous, right?
[35:59.540 --> 36:05.940] [SPEAKER_01]: Which is that the data that is being generated by these devices can be trusted.
[36:07.000 --> 36:11.320] [SPEAKER_01]: What if we made the data that was generated by these devices untrustworthy?
[36:12.060 --> 36:14.480] [SPEAKER_01]: What if we could generate nonsense data?
[36:14.660 --> 36:25.060] [SPEAKER_01]: What if we could generate crap data that completely obfuscated our actual viewing patterns and made that data completely useless from an advertising perspective?
[36:25.280 --> 36:29.700] [SPEAKER_01]: Like, this is something I've always been thinking about for, you know, 15 years or so.
[36:29.840 --> 36:39.460] [SPEAKER_01]: Like, why is there not some kind of filter or some kind of device that can continually generate crap data so that this type of system does not exist anymore?
[36:39.820 --> 36:43.480] [SPEAKER_01]: So that this type of system can essentially be broken by the consumer.
[36:44.060 --> 36:44.840] [SPEAKER_03]: Jamming the radar.
[36:45.180 --> 36:45.320] [SPEAKER_03]: Yeah.
[36:45.760 --> 36:47.460] [SPEAKER_01]: Yeah, let's jam the radar.
[36:47.680 --> 36:48.200] [SPEAKER_01]: I mean, why not?
[36:48.320 --> 36:49.760] [SPEAKER_01]: I mean, look, we're hackers, right?
[36:49.860 --> 36:53.280] [SPEAKER_01]: We should be able to generate random data on a particular device.
[36:53.520 --> 37:00.320] [SPEAKER_01]: I don't know if it would require flashing the firmware, but that's not something that is too difficult for us to do.
[37:00.460 --> 37:03.860] [SPEAKER_01]: We've been playing with EEPROM reader writers since the early 90s.
[37:03.860 --> 37:05.600] [SPEAKER_01]: I mean, let's get to work here.
[37:05.780 --> 37:14.100] [SPEAKER_01]: Like, I would love to see somebody build something like this for the greater good here and talk about it at the next Hope Conference.
[37:14.340 --> 37:15.540] [SPEAKER_00]: It would be amazing.
[37:15.760 --> 37:29.340] [SPEAKER_00]: I mean, there's a huge scene out there for people who want to do this sort of thing with their smartphones, install alternative OSs, or at least, you know, root the thing, gain more control over what you are and aren't allowed to do with it by the operating system.
[37:30.180 --> 37:32.000] [SPEAKER_00]: And smartphones, they're ubiquitous.
[37:32.000 --> 37:34.880] [SPEAKER_00]: Yes, everyone has one, but smart TVs are getting there.
[37:35.020 --> 37:40.740] [SPEAKER_00]: And there does not seem to be as an equivalent scene out there for hacking these things.
[37:40.920 --> 37:51.300] [SPEAKER_00]: And that's something that I think we would love to see because, you know, if you want to get the smart TV, you don't want to pay the extra for the dumb TV, but you want to subvert how it works.
[37:51.300 --> 37:54.660] [SPEAKER_00]: You know, can I cut things off at the firewall?
[37:54.920 --> 37:57.000] [SPEAKER_00]: Can I, you know, put in...
[37:57.520 --> 38:05.160] [SPEAKER_00]: How great would it be to put an open-source operating system on your television, making it into a computer that you could basically do whatever you want with?
[38:05.680 --> 38:12.080] [SPEAKER_00]: There are so many possibilities for things, and it needs brilliant minds to start hacking away at the problem.
[38:13.900 --> 38:15.640] [SPEAKER_01]: Yeah, I completely agree.
[38:15.860 --> 38:25.120] [SPEAKER_01]: I mean, and for somebody like you, it would probably be really important to try to obfuscate your viewing habits so that people don't realize you're such a big friends fan.
[38:27.100 --> 38:29.640] [SPEAKER_01]: And yeah, to go back to that once again.
[38:30.240 --> 38:32.800] [SPEAKER_01]: So there's a vested interest in this on your part.
[38:33.540 --> 38:46.260] [SPEAKER_00]: Yeah, just I would love someone in some crapware company somewhere looking at the results and going, this guy is just watching a lot of watching the same episode of men behaving badly from the 90s over and over 24-7.
[38:46.800 --> 38:49.160] [SPEAKER_04]: We really like Rob Schneider in this house.
[38:50.220 --> 38:51.460] [SPEAKER_04]: We don't actually.
[38:52.420 --> 38:53.020] [SPEAKER_00]: Yes.
[38:54.180 --> 38:55.320] [SPEAKER_04]: Ron Eldard.
[38:55.560 --> 38:56.460] [SPEAKER_00]: Ron Eldard.
[38:56.680 --> 38:58.040] [SPEAKER_04]: And I hate that I know that.
[39:01.010 --> 39:01.950] [SPEAKER_00]: Fun stuff.
[39:03.830 --> 39:06.150] [SPEAKER_00]: What else is going on?
[39:06.270 --> 39:25.730] [SPEAKER_00]: There is there just this past the past few days have been rattling around with the news that Firefox, the the head of Mozilla, the the new head, the new CEO of Mozilla, Anthony Enzor DeMeo, has announced that Firefox is going to be a modern AI browser.
[39:27.050 --> 39:32.930] [SPEAKER_00]: Firefox, of course, a very popular open-source web browser that has been, you know, has had ebbs and flows.
[39:32.950 --> 39:34.010] [SPEAKER_00]: It used to be really huge.
[39:34.110 --> 39:39.810] [SPEAKER_00]: Now it's still got a dedicated user base that it seems intent on making angry.
[39:41.370 --> 39:48.750] [SPEAKER_00]: But, you know, there have been AI features that have been trickled onto the thing that are relatively easy to disable.
[39:49.150 --> 39:55.750] [SPEAKER_00]: But people are questioning now if this is indicative of a direction that the organization is taking.
[39:56.250 --> 39:57.970] [SPEAKER_04]: I have a question.
[39:58.310 --> 39:58.490] [SPEAKER_04]: Yes.
[39:58.590 --> 40:05.370] [SPEAKER_04]: And that's something that has bothered me with so much of this with co-pilot on your TV with AI in your browser.
[40:05.530 --> 40:06.470] [SPEAKER_04]: To what end?
[40:08.110 --> 40:14.610] [SPEAKER_00]: To the end that I think as we were talking about it's ways for the people behind these things to continue to monetize you.
[40:16.690 --> 40:27.810] [SPEAKER_00]: the old model of just going and buying a piece of equipment and appliance and anything and that being the financial transaction then you take the thing home and you have the thing and you can use the thing because it's your thing.
[40:29.230 --> 40:31.150] [SPEAKER_00]: But companies like money.
[40:31.310 --> 40:32.550] [SPEAKER_00]: They want more of yours.
[40:32.850 --> 40:39.510] [SPEAKER_00]: And if they could continue to monetize your use of the product after they've already sold it to you, well, let's get on that.
[40:39.750 --> 40:40.990] [SPEAKER_00]: You know, the shareholders would love it.
[40:44.710 --> 40:49.810] [SPEAKER_00]: And it is something which I think is very concerning.
[40:50.150 --> 40:51.450] [SPEAKER_00]: It should be of more concern.
[40:51.710 --> 41:06.210] [SPEAKER_00]: Not enough people are talking about it, but with the current AI bubble looming, I think there is maybe something in the hope that we can maybe get more folks angry about it.
[41:06.630 --> 41:10.890] [SPEAKER_00]: You know, first you've got to get mad, as a man once said.
[41:11.310 --> 41:11.850] [SPEAKER_00]: Yes, Alex.
[41:12.810 --> 41:22.790] [SPEAKER_01]: Well, it's an interesting pivot from Firefox's perspective to have some kind of AI enabled browser functionality.
[41:23.010 --> 41:34.190] [SPEAKER_01]: I don't necessarily even know what the hell it means to be like an AI browser or something, but presumably it's like with a bunch of annoying AI tools.
[41:34.810 --> 41:38.350] [SPEAKER_01]: But on the other hand, it does make some sense.
[41:38.650 --> 41:57.310] [SPEAKER_01]: And if you think about how things have changed, even just over the last three years or so, really gone are the days when you would go to a search engine like Google, type in a question, and then webpages would show up, and then you would have to read those webpages and answer the question for yourself.
[41:57.730 --> 42:15.470] [SPEAKER_01]: Now, what happens on all of the search engines is that an AI system will already pre-read all of those results for you and have some kind of summary of whatever the answer is to the question that you post that you no longer even have to click on these particular websites anymore.
[42:16.010 --> 42:23.550] [SPEAKER_01]: And it's like the fact that websites exist is somewhat deprecated already, right?
[42:23.650 --> 42:35.410] [SPEAKER_01]: Because the information exists in the LLM or whatever the generative AI system is, and it's spitting out the answer so that you don't even need these websites anymore.
[42:35.410 --> 42:50.690] [SPEAKER_01]: are so to have a web or an AI-enabled browser for things that you don't even necessarily need to read seems to me like just another interface into LLM systems or AI-generated content in the first instance.
[42:51.030 --> 43:03.990] [SPEAKER_01]: So what is the difference then between the new Firefox client, the new Firefox web browser that's AI-enabled, and just going into ChatGPT or perplexity or something and asking the same question?
[43:04.950 --> 43:10.390] [SPEAKER_01]: Functionally, I'd really like to know, because it's a weird world in which we're living now.
[43:11.430 --> 43:16.410] [SPEAKER_00]: It is, and that relates to another story which has been going around, which is about Amazon Prime Video.
[43:18.050 --> 43:27.290] [SPEAKER_00]: Producers of various video content, TV shows, things like that, they decided to add a little feature which would give you an AI-powered video recap if you were watching a series.
[43:27.290 --> 43:31.950] [SPEAKER_00]: It would tell you what happened on previous episodes in case you needed little reminders.
[43:31.950 --> 43:37.470] [SPEAKER_00]: This was on the show's Fallout, The Rig, Tom Clancy's Jack Ryan, Upload, and Bosh.
[43:38.570 --> 43:47.650] [SPEAKER_00]: The feature is supposed to use AI, this story from The Verge, to analyze a show's key plot points and sum it all up in a bite-sized video complete with an AI voiceover and clips from the series.
[43:48.210 --> 43:49.290] [SPEAKER_00]: They tried this.
[43:50.050 --> 43:51.890] [SPEAKER_00]: It failed spectacularly.
[43:52.230 --> 43:55.990] [SPEAKER_00]: The little blurbs that were generated were getting facts of the shows wrong.
[43:56.210 --> 43:59.650] [SPEAKER_00]: They were getting plot points wrong.
[43:59.790 --> 44:02.890] [SPEAKER_00]: They were getting all sorts of things completely incorrect.
[44:07.490 --> 44:14.130] [SPEAKER_00]: The customers roundly laughed at it, mocked it, and Amazon quietly pulled the feature.
[44:15.090 --> 44:22.330] [SPEAKER_00]: But even a company on the scale of Amazon with the powerful machinery behind that could not get something like this working.
[44:22.550 --> 44:23.930] [SPEAKER_00]: They pushed it to production anyway.
[44:24.130 --> 44:29.870] [SPEAKER_00]: They put it out there and it failed hard because that's what people are doing.
[44:29.990 --> 44:33.230] [SPEAKER_00]: That's what the current technological environment...
[44:33.230 --> 44:39.030] [SPEAKER_00]: You've got to have AI something, you've got to push it out there, you've got to shove it in front of people, whether or not they asked for it.
[44:39.150 --> 44:40.170] [SPEAKER_04]: And whether or not it works.
[44:40.370 --> 44:41.850] [SPEAKER_00]: And whether or not it works.
[44:42.170 --> 44:45.010] [SPEAKER_04]: I'm curious what the tipping point is going to have to be.
[44:45.650 --> 44:53.330] [SPEAKER_04]: Because there's so much that doesn't work and is being pushed out in ways that people don't want, don't appreciate, feel they're stuck with.
[44:53.330 --> 44:56.770] [SPEAKER_04]: And then it's wrong on so many levels.
[44:57.150 --> 45:03.230] [SPEAKER_04]: And in the same way that, you know, I don't know, three, four years ago, NFTs were the next hot thing.
[45:03.490 --> 45:05.850] [SPEAKER_04]: And then everyone realized, wait, they were actually really stupid.
[45:07.390 --> 45:11.870] [SPEAKER_04]: How long is it going to take to get us to the AI tipping point?
[45:11.990 --> 45:12.770] [SPEAKER_04]: And will there be one?
[45:12.770 --> 45:12.770] Right.
[45:13.190 --> 45:25.450] [SPEAKER_00]: And that's the thing, you know, with the example you were just giving, Alex, of, you know, can you have an AI browser that just sums up the information out there and feeds it to you without you even having to go through the tedium of loading a website and reading it with your eyes?
[45:27.010 --> 45:29.790] [SPEAKER_00]: The tech is not good at getting things correct.
[45:30.270 --> 45:34.510] [SPEAKER_00]: It's very good at confidently being wrong, like some folks we know.
[45:34.670 --> 45:38.490] [SPEAKER_00]: But it is a tool that is not ready for prime time.
[45:38.490 --> 45:40.470] [SPEAKER_00]: It doesn't do the things that it's labeled to do.
[45:40.750 --> 45:46.330] [SPEAKER_00]: And people are forcing it out there in any in ways that even if they worked, it wouldn't make sense.
[45:46.450 --> 45:47.950] [SPEAKER_00]: But the fact is, they don't.
[45:48.410 --> 45:50.290] [SPEAKER_00]: And this is it is fascinating.
[45:50.430 --> 46:03.030] [SPEAKER_00]: It is going to be really fascinating to see how this all shapes up when, you know, the the sheer demand for RAM has driven the price way up to fuel all these things.
[46:04.390 --> 46:11.510] [SPEAKER_00]: And companies still are shoving it in front of you and hoping that you just accept that this is the way things work now.
[46:12.030 --> 46:15.090] [SPEAKER_00]: Search engines are garbage now compared to where they were six, seven years ago.
[46:16.790 --> 46:19.710] [SPEAKER_00]: Television is now watching itself to save you the trouble.
[46:21.870 --> 46:24.790] [SPEAKER_00]: And I'm just I'm just waiting for an electric monk to turn up.
[46:26.110 --> 46:28.230] [SPEAKER_00]: Douglas Adams fans will know what that's all about.
[46:28.450 --> 46:36.930] [SPEAKER_00]: But it's, it's a it's a fascinating time to be involved in this to be a consumer to be somebody who is keeping an eye on the tech side of it all.
[46:37.390 --> 46:41.130] [SPEAKER_00]: And we're going to keep doing our best here to continue to do that.
[46:41.330 --> 46:45.270] [SPEAKER_00]: And on that note, you out there can help us continue to do that.
[46:45.570 --> 46:48.750] [SPEAKER_00]: The WBAI fundraiser is ongoing.
[46:49.570 --> 46:54.290] [SPEAKER_00]: We are really on track to meet our goal by the end of the year, but we need your help to do that.
[46:54.470 --> 47:01.550] [SPEAKER_00]: Go to buddy.wbai.org or give us a call at 833-WBAI-NYC.
[47:01.550 --> 47:04.550] [SPEAKER_00]: That's 833-922-4692.
[47:04.990 --> 47:16.150] [SPEAKER_00]: Or go to buddy.wbai.org with a browser that works and throw in what support you can for our home in the heart of community radio in the heart of the FM dial.
[47:16.610 --> 47:19.670] [SPEAKER_00]: Keep us doing what we do here on this station here on this program.
[47:20.450 --> 47:25.630] [SPEAKER_00]: Keep, keep your support coming because, you know, we're just about out of 2025.
[47:26.170 --> 47:29.150] [SPEAKER_00]: How 2026 starts for us is up to you.
[47:30.050 --> 47:36.870] [SPEAKER_00]: You could give a one time donation, you could become a WBAI buddy with a recurring donation, which we really appreciate, but every little bit helps.
[47:36.870 --> 47:37.370] [SPEAKER_04]: It does.
[47:37.450 --> 47:40.530] [SPEAKER_04]: And if they ask you what show you're supporting, please mention us.
[47:42.070 --> 47:46.230] [SPEAKER_04]: It, you know, it helps the powers that be know that we have a strong listening audience.
[47:46.430 --> 47:47.110] [SPEAKER_04]: It's hugely helpful.
[47:48.470 --> 47:58.230] [SPEAKER_04]: And yeah, also, please remember that WBAI and Pacifica Foundation are a 501c3 organization and your donations may in fact be tax deductible.
[47:58.230 --> 48:06.930] [SPEAKER_04]: If you are a person who is required to make a required minimum distribution from your IRA before the end of this year, WBAI is a great place to do it.
[48:07.590 --> 48:08.890] [SPEAKER_00]: And you still got a little time.
[48:09.510 --> 48:12.810] [SPEAKER_04]: Check that out with your financial advisors, but it's a good thing to do.
[48:12.950 --> 48:13.930] [SPEAKER_04]: It's a good way to help.
[48:13.990 --> 48:20.210] [SPEAKER_04]: And it keeps the station up and running in ways that would not be possible otherwise.
[48:20.210 --> 48:31.670] [SPEAKER_00]: And with the media environment being an increasing mess, let's just point out that everything we do on this program is human-generated by us people who do this because we care.
[48:32.050 --> 48:36.830] [SPEAKER_00]: And we appreciate you listening and helping us along with it.
[48:38.810 --> 48:41.310] [SPEAKER_00]: We are nearly out of time.
[48:42.590 --> 48:47.410] [SPEAKER_00]: Just remind folks, send us email, oth@2600.com.
[48:47.850 --> 48:50.570] [SPEAKER_00]: A few of you out there, we always appreciate hearing from you.
[48:50.690 --> 48:53.210] [SPEAKER_00]: Even if we don't read the notes on the air, we get them.
[48:53.910 --> 48:57.470] [SPEAKER_00]: A few people have been sending us photos of payphones for 2600 Magazine.
[48:57.690 --> 48:59.230] [SPEAKER_00]: This is not the address to do that.
[48:59.350 --> 49:00.950] [SPEAKER_00]: That's the address from the magazine.
[49:01.290 --> 49:07.350] [SPEAKER_00]: But we appreciate you nonetheless, oth@2600.com.
[49:07.350 --> 49:21.730] [SPEAKER_00]: And that'll have to do it for this week for our absent friends, Emmanuel and Kyle, for folks everywhere, for Yila, for me, for our WBAI family.
[49:22.070 --> 49:25.490] [SPEAKER_00]: We will be back next week with another edition of Off The Hook.
[49:26.890 --> 49:28.390] [SPEAKER_00]: Take care of yourselves out there.
[49:30.170 --> 49:31.290] [SPEAKER_00]: Keep your eyes open.
[49:34.110 --> 49:34.990] [SPEAKER_00]: Keep listening.
[49:38.290 --> 49:39.350] [SPEAKER_00]: And have a good night.