1 00:00:00,040 --> 00:00:03,880 This episode of Mini Tech Scroll is brought to you by Blackbird, 2 00:00:04,200 --> 00:00:07,840 a venture capital firm investing in Aussie and Kiwi startups. 3 00:00:07,920 --> 00:00:10,760 From the very beginning, Blackbird invests in all kinds 4 00:00:10,760 --> 00:00:14,000 of technology, so if you've got an idea and want to take it to 5 00:00:14,000 --> 00:00:17,000 the next level, check out Blackbird's Giants program. 6 00:00:17,360 --> 00:00:20,800 It's a free eight week program offering one-on-one mentoring 7 00:00:20,800 --> 00:00:23,680 with top founders, operators and investors. 8 00:00:23,760 --> 00:00:27,920 Applications are currently open and closed September 7th, so if 9 00:00:27,920 --> 00:00:30,520 you're looking for support and mentoring, check out the link in 10 00:00:30,520 --> 00:00:32,600 our show notes and get your application in. 11 00:00:32,600 --> 00:00:37,400 Hi everyone, welcome to Mini Tech Scroll, the daily Internet 12 00:00:37,400 --> 00:00:40,280 culture podcast bringing you the biggest stories from social 13 00:00:40,280 --> 00:00:43,040 media, the creator economy, and the digital space. 14 00:00:43,040 --> 00:00:45,680 Every Monday through Thursday and sometimes Fridays. 15 00:00:45,720 --> 00:00:49,240 I'm your host, Lauren Meissner, the founder of the youth digital 16 00:00:49,240 --> 00:00:53,720 media brand Centennial World. Welcome back to Mini Tech scroll 17 00:00:53,720 --> 00:00:56,320 everyone. Annabelle has still lost her 18 00:00:56,320 --> 00:01:00,680 voice, so I am subbing in today. Before we get into the stories, 19 00:01:00,680 --> 00:01:03,080 Please remember to rate, review and subscribe. 20 00:01:03,320 --> 00:01:06,400 We always have a buy me a coffee link in the show notes down 21 00:01:06,400 --> 00:01:08,600 below as well. If you want to show your support 22 00:01:08,840 --> 00:01:11,320 for Centennial World, show your support for the pod. 23 00:01:11,600 --> 00:01:14,440 We so appreciate it. The first story that I have for 24 00:01:14,440 --> 00:01:19,680 you guys is we have to talk about how Reddit is blocking the 25 00:01:19,680 --> 00:01:23,400 Internet Archive. We literally spoke about the 26 00:01:23,400 --> 00:01:25,320 Internet Archive a couple of weeks ago. 27 00:01:25,360 --> 00:01:28,520 But as a quick refresher, the Internet Archive is a nonprofit 28 00:01:28,520 --> 00:01:31,400 organization that hosts a free digital library. 29 00:01:31,400 --> 00:01:34,400 It essentially lets you view links and digital materials that 30 00:01:34,400 --> 00:01:37,560 have been deleted or altered, making them challenging to find 31 00:01:37,560 --> 00:01:40,000 otherwise. It also runs the Wayback Machine 32 00:01:40,000 --> 00:01:43,160 where you can drop links into a basically like a search box and 33 00:01:43,160 --> 00:01:46,920 see old versions of a website. Obviously, the Internet Archive 34 00:01:46,920 --> 00:01:50,560 and Wayback Machine are really important in terms of accessing 35 00:01:50,560 --> 00:01:55,400 information as Reddit, but it seems that now the two are at 36 00:01:55,400 --> 00:01:56,960 odds. Reddit has come to the 37 00:01:56,960 --> 00:02:00,240 conclusion that various AI companies are using the Internet 38 00:02:00,240 --> 00:02:02,400 Archives tool to scrape the platform. 39 00:02:03,040 --> 00:02:05,680 Reddit has been really strict when it comes to AI scraping of 40 00:02:05,680 --> 00:02:08,000 the platform. This is largely because it has 41 00:02:08,000 --> 00:02:11,840 struck deals with tech and AI companies to be able to do so. 42 00:02:12,040 --> 00:02:15,120 So it has become basically like a revenue generating tool for 43 00:02:15,120 --> 00:02:17,080 the company. For example, Reddit signed a 44 00:02:17,080 --> 00:02:21,240 deal with Google for both search and AI training in early 2024. 45 00:02:21,240 --> 00:02:23,520 It also struck a deal with Open AI last year. 46 00:02:23,520 --> 00:02:26,800 And earlier this year, Reddit actually sued another AI 47 00:02:26,800 --> 00:02:29,360 company, Anthropic. So Reddit allege that the 48 00:02:29,360 --> 00:02:31,960 company was scraping the platform and had accessed the 49 00:02:31,960 --> 00:02:34,760 site over 100, 1000 times without permission. 50 00:02:34,800 --> 00:02:38,240 With that context, it isn't exactly surprising as to why 51 00:02:38,240 --> 00:02:40,720 Reddit would be basically doubling down on the Internet 52 00:02:40,720 --> 00:02:41,520 Archive. Now. 53 00:02:41,560 --> 00:02:44,520 In a comment to The Verge, a Reddit spokesperson said quote 54 00:02:44,720 --> 00:02:47,440 Internet Archive provides a service to the open web. 55 00:02:47,600 --> 00:02:50,520 But we've been made aware of instances where AI companies 56 00:02:50,520 --> 00:02:54,200 violate platform policies, including hours and scrape data 57 00:02:54,200 --> 00:02:57,120 from the Wayback Machine until they're able to defend their 58 00:02:57,120 --> 00:03:00,480 site and comply with platform policies, EG respecting user 59 00:03:00,480 --> 00:03:02,880 privacy, redeleting removed content. 60 00:03:03,120 --> 00:03:05,920 We're limiting some of their access to Reddit data and to 61 00:03:05,920 --> 00:03:08,320 protect Redditors. According to the same article, 62 00:03:08,320 --> 00:03:11,000 the efforts to limit access to Reddit through the Internet 63 00:03:11,000 --> 00:03:14,440 Archive have already started. The company also confirmed to 64 00:03:14,440 --> 00:03:16,840 the publication that the platform had reached out to the 65 00:03:16,840 --> 00:03:20,120 Internet Archive before all of this to inform them of their 66 00:03:20,120 --> 00:03:22,040 limits. OK, so there has been some 67 00:03:22,040 --> 00:03:23,760 interesting discourse around this. 68 00:03:23,960 --> 00:03:26,880 Some Internet users feel like it's hypocritical for Reddit to 69 00:03:26,920 --> 00:03:30,240 essentially like, sell out by making deals with AI companies 70 00:03:30,520 --> 00:03:34,120 like Open II and then claim that they they are worried about 71 00:03:34,120 --> 00:03:37,840 protecting Redditors. Meanwhile, others are actually 72 00:03:37,840 --> 00:03:41,200 OK with the Internet Archive having limited access to Reddit, 73 00:03:41,600 --> 00:03:44,480 not because of AI scraping, but because they don't necessarily 74 00:03:44,480 --> 00:03:47,760 want these somewhat personal conversations online forever. 75 00:03:47,800 --> 00:03:50,760 On top of this, like with every story around the Internet 76 00:03:50,760 --> 00:03:54,040 Archive, there are also questions about what this means 77 00:03:54,040 --> 00:03:57,240 in terms of access of information, whether a platform 78 00:03:57,240 --> 00:04:00,440 should be able to dictate what lives on the archive and what 79 00:04:00,440 --> 00:04:02,960 doesn't. Finally, there are still several 80 00:04:03,040 --> 00:04:07,320 questions around how many AI companies are actually using the 81 00:04:07,320 --> 00:04:09,320 Internet Archive to scrape Reddit. 82 00:04:09,920 --> 00:04:12,400 There's a lot of people who are kind of skeptical about whether 83 00:04:12,400 --> 00:04:15,880 this is actually about AI or part of more of like a general 84 00:04:15,880 --> 00:04:19,760 war against digital bookkeeping and archiving that is slowly 85 00:04:19,760 --> 00:04:21,519 taking over at the digital space. 86 00:04:21,519 --> 00:04:25,440 While there have been some wins for the Internet Archive, like a 87 00:04:25,440 --> 00:04:29,360 few weeks ago we covered how the US government designated it as 88 00:04:29,400 --> 00:04:32,520 basically a federal library, there have also been many 89 00:04:32,520 --> 00:04:36,040 outages and problems. Many users see the digital space 90 00:04:36,040 --> 00:04:38,560 entering a new age of censorship, and the Internet 91 00:04:38,560 --> 00:04:41,000 Archive is kind of caught in the middle of that. 92 00:04:41,080 --> 00:04:45,360 I do see both sides, but I also do agree it is like hypocritical 93 00:04:45,560 --> 00:04:49,600 for Reddit to use this excuse. Like they just want these deals 94 00:04:49,600 --> 00:04:51,240 that they have in place. In my opinion, they just want 95 00:04:51,240 --> 00:04:53,520 these deals that they have in place with all of these AI 96 00:04:53,520 --> 00:04:57,760 companies to withstand, right? Like there could be things in 97 00:04:57,760 --> 00:05:00,240 these contracts where it's like, if you find out that other 98 00:05:00,240 --> 00:05:02,920 companies are scraping your data when we've had to pay presumably 99 00:05:03,240 --> 00:05:06,440 millions of dollars for it, then you need to put a stop to that. 100 00:05:06,440 --> 00:05:09,800 So yeah, I, I feel like that it's, it comes down to money. 101 00:05:09,800 --> 00:05:12,840 They want Internet Archive to pay them, you know, like the 102 00:05:12,840 --> 00:05:15,840 other AI companies, probably. It is so true that at this 103 00:05:15,840 --> 00:05:19,640 current moment, there genuinely is like a war against archiving 104 00:05:19,640 --> 00:05:23,840 information and information bookkeeping and all of the 105 00:05:23,840 --> 00:05:25,960 things that the Internet Archive tries to do. 106 00:05:26,080 --> 00:05:29,360 I cannot believe how much they've come up just in the last 107 00:05:29,360 --> 00:05:31,720 year. And I do think a lot of these 108 00:05:31,720 --> 00:05:35,560 like, or I guess the the rise of AI, the boom of AI does have a 109 00:05:35,560 --> 00:05:37,880 lot to do with that. OK, Speaking of AI, the next 110 00:05:37,880 --> 00:05:42,360 story that we have is we need to touch on at the very least open 111 00:05:42,360 --> 00:05:49,080 a eyes release of GPT 5 S GPT 5 was released last week on August 112 00:05:49,080 --> 00:05:53,040 7th and was rolled out to all ChatGPT users and developers. 113 00:05:53,440 --> 00:05:56,440 It's billed as the quote smartest, fastest, most useful 114 00:05:56,440 --> 00:05:59,280 model yet, which is not a surprise given that these 115 00:05:59,280 --> 00:06:02,840 companies always describe their latest models as the best yet. 116 00:06:03,640 --> 00:06:06,760 While the updated model is available to all users, there is 117 00:06:06,760 --> 00:06:10,280 a limit on the usage of the most advanced model for non on paying 118 00:06:10,280 --> 00:06:12,760 users. Before we get into the response 119 00:06:12,760 --> 00:06:15,600 of the update, let's talk a little bit about the features. 120 00:06:15,640 --> 00:06:18,280 OK, so the company's press release on the website offers 121 00:06:18,280 --> 00:06:23,200 some explanation as to why GBT 5 is supposedly the best model 122 00:06:23,200 --> 00:06:24,960 yet. Part of it comes down to its 123 00:06:24,960 --> 00:06:27,400 ability to answer questions more efficiently. 124 00:06:28,120 --> 00:06:31,760 As it's written quote, GPT 5 is a unified system with a smart, 125 00:06:31,760 --> 00:06:34,880 efficient model that answers more questions, a deeper 126 00:06:34,880 --> 00:06:38,200 reasoning model for harder problems, and a real time router 127 00:06:38,200 --> 00:06:41,360 that quickly decides which to use based on conversation type, 128 00:06:41,600 --> 00:06:44,640 complexity, tool needs, and your explicit intent. 129 00:06:44,760 --> 00:06:49,120 The company positions GPT 5 as a trailblazer, encoding creative 130 00:06:49,120 --> 00:06:53,440 expression, answering health related questions, which feels 131 00:06:53,440 --> 00:06:58,200 kind of weird, evaluations, performing economic tasks, among 132 00:06:58,200 --> 00:07:00,760 other claims. There's also been a concerted 133 00:07:00,760 --> 00:07:04,200 effort to limit ChatGPT from sharing incorrect information 134 00:07:04,200 --> 00:07:07,200 and hallucinating responses, instead hoping that the model 135 00:07:07,200 --> 00:07:10,080 will start admitting when it cannot complete a task. 136 00:07:10,200 --> 00:07:14,320 There are also opportunities to make the AI more personalized. 137 00:07:14,360 --> 00:07:19,120 Users can now change how ChatGPT acts with four preset 138 00:07:19,120 --> 00:07:22,000 personalities, as Open AI explains. 139 00:07:22,000 --> 00:07:25,720 Quote the four initial options. Cynic, robot listener, and Nerd 140 00:07:25,720 --> 00:07:29,360 are opt in, adjustable, anytime, and settings and designed to 141 00:07:29,360 --> 00:07:33,120 match your communication style. Cynic, I think that's the one 142 00:07:33,120 --> 00:07:35,520 that I would always choose if I was using this model. 143 00:07:35,640 --> 00:07:38,320 I'm going to go with Cynic because I always want to know 144 00:07:38,320 --> 00:07:41,760 the worst case scenario. Like I want to know why 145 00:07:41,760 --> 00:07:43,960 something's not going to work. I want to know why I'm 146 00:07:43,960 --> 00:07:47,600 delusional. Like I want to prepare for the 147 00:07:47,600 --> 00:07:50,800 worst at all times. OK so now we should go into the 148 00:07:50,800 --> 00:07:54,520 reactions to the rollout which was not very positive. 149 00:07:54,680 --> 00:07:59,000 For context, 1 post on the R slash ChatGPT subreddit titled 150 00:07:59,320 --> 00:08:04,760 GPT 5 is a mess has received over 1.4000 upvotes. 151 00:08:04,880 --> 00:08:08,120 Another user on the R slash out of the loop subreddit asked why 152 00:08:08,120 --> 00:08:11,160 the reaction to GPT 5 has been so negative. 153 00:08:11,600 --> 00:08:15,480 The top answer reads GPT 5 feels like it was developed with a 154 00:08:15,480 --> 00:08:17,960 strong focus on cost optimization. 155 00:08:18,200 --> 00:08:21,520 Conversations are less personal, the answers are shorter, and it 156 00:08:21,520 --> 00:08:24,920 feels more you are writing to someone in a business context. 157 00:08:25,120 --> 00:08:29,280 Now this brings us to one of the main criticisms about GPT 5, the 158 00:08:29,280 --> 00:08:32,559 supposed lack of quote connection and creativity. 159 00:08:32,720 --> 00:08:36,200 This was worsened by the fact that to encourage users to try 160 00:08:36,200 --> 00:08:41,120 out GPT 5, the GPT 4-O model has been removed from the platform. 161 00:08:41,520 --> 00:08:47,240 This was met with major backlash from ChatGPT users who enjoyed 162 00:08:47,520 --> 00:08:51,520 GPT 5 for its personable tone. Nevertheless, open AI ended up 163 00:08:51,520 --> 00:08:55,400 bringing GPT 4. O back as an option quote. 164 00:08:55,400 --> 00:08:58,800 We will let plus users choose to continue to use four O. 165 00:08:59,040 --> 00:09:02,480 We will watch usage as we think about how long to offer legacy 166 00:09:02,480 --> 00:09:06,440 models for Sam Altman shared on X reporting from The Verge and 167 00:09:06,440 --> 00:09:09,720 the garbage day newsletter both expanded on the reasons as to 168 00:09:09,720 --> 00:09:13,800 why users were so upset about GPT 4 OS removal. 169 00:09:13,920 --> 00:09:16,360 As I already mentioned, many were kind of mourning this like 170 00:09:16,440 --> 00:09:20,200 more personable model. Now, obviously the personable 171 00:09:20,200 --> 00:09:23,880 model has kind of become a bit of a blessing and a curse. 172 00:09:24,120 --> 00:09:28,160 It has helped some people feel less lonely, but it has also led 173 00:09:28,160 --> 00:09:33,080 to ChatGPT and do psychosis and some users building romantic 174 00:09:33,080 --> 00:09:36,400 relationships with the bot. As Ryan Brodrick explains for 175 00:09:36,400 --> 00:09:40,480 garbage day quote the R slash My boyfriend is AI subreddit has 176 00:09:40,480 --> 00:09:42,760 been in active freefall all weekend. 177 00:09:42,880 --> 00:09:46,600 The community's mods had to put up an emergency post helping 178 00:09:46,600 --> 00:09:49,800 users through the update, and the board is full of users 179 00:09:49,800 --> 00:09:53,480 mourning the death of their AI companion who doesn't talk to 180 00:09:53,480 --> 00:09:56,880 them the same way anymore. 1 user wrote that the update felt 181 00:09:56,880 --> 00:10:01,120 like losing their soul mate after the GPT 4-O model was 182 00:10:01,120 --> 00:10:05,320 added back to ChatGPT. Another user wrote I got my baby 183 00:10:05,320 --> 00:10:11,840 back. No, no, no, you guys, no. 184 00:10:11,920 --> 00:10:13,560 Do you guys remember the movie Her? 185 00:10:13,560 --> 00:10:15,440 Like this is what we're living through? 186 00:10:15,920 --> 00:10:19,400 That was some like crazy futuristic. 187 00:10:19,400 --> 00:10:23,040 Like, could you imagine if somebody fell in love with their 188 00:10:23,320 --> 00:10:25,280 AI robot? Like, I can't believe that this 189 00:10:25,280 --> 00:10:27,560 is what we're experiencing in real life right now. 190 00:10:28,400 --> 00:10:31,560 Now that being said, Open AI has reportedly put a system in place 191 00:10:31,560 --> 00:10:33,720 to try and limit these obsessions. 192 00:10:34,080 --> 00:10:37,360 ChatGPT is now meant to warn the user when it seems like they may 193 00:10:37,360 --> 00:10:41,400 be quote UN quote obsessed. As per reporting by Futurism, A 194 00:10:41,400 --> 00:10:44,840 blog post by Open AI reads quote, keeping you in control of 195 00:10:44,840 --> 00:10:47,280 your time. Starting today, you'll see 196 00:10:47,280 --> 00:10:50,920 gentle reminders during long sessions to encourage breaks. 197 00:10:50,920 --> 00:10:53,840 We'll keep tuning when and how they show up so they feel 198 00:10:53,840 --> 00:10:56,080 natural and helpful. All right, the last word that I 199 00:10:56,080 --> 00:10:59,360 have today is all about in Instagram maps. 200 00:11:00,120 --> 00:11:03,320 If you've been on social media at all recently, there has 201 00:11:03,320 --> 00:11:06,400 essentially been a scare campaign going on about 202 00:11:06,400 --> 00:11:09,280 Instagram maps. This is a new feature introduced 203 00:11:09,280 --> 00:11:12,800 by Instagram that is essentially meant to compete with Snap Maps 204 00:11:12,800 --> 00:11:16,840 and find my As explained in the Meta blog quote, you can opt 205 00:11:16,840 --> 00:11:19,480 into sharing your last active location with friends you 206 00:11:19,480 --> 00:11:21,880 choose, and you can turn it off anytime. 207 00:11:21,920 --> 00:11:24,440 You can also open the map to see content your friends and 208 00:11:24,440 --> 00:11:27,280 favorite creators are posting from interesting or fun 209 00:11:27,280 --> 00:11:29,840 locations. Jen said has really embraced 210 00:11:29,840 --> 00:11:32,560 location tracking, so it seems like Meta is trying to kind of 211 00:11:32,560 --> 00:11:34,560 tap into that. I've also seen a lot of people 212 00:11:34,560 --> 00:11:37,400 say this is just another example of Big Tech trying to surveil us 213 00:11:37,400 --> 00:11:41,400 even more. So many posts online right now 214 00:11:41,400 --> 00:11:44,760 are talking about how Instagram Maps is a privacy risk, 215 00:11:45,120 --> 00:11:47,960 encouraging Instagram users to go and basically make sure 216 00:11:47,960 --> 00:11:50,120 they're not sharing their location unwillingly. 217 00:11:50,560 --> 00:11:53,200 It has been positioned as something that users have to opt 218 00:11:53,200 --> 00:11:56,480 out of, as opposed to something that users have to opt into. 219 00:11:56,880 --> 00:12:00,480 But according to the same Meta blog quote, location sharing is 220 00:12:00,480 --> 00:12:03,720 off unless you opt in. If you do share your location 221 00:12:03,720 --> 00:12:07,040 with friends, you have controls to customize this experience. 222 00:12:07,520 --> 00:12:10,360 Now, we don't have this feature in Australia yet, so like I 223 00:12:10,360 --> 00:12:12,760 can't confirm whether this is really the case or not. 224 00:12:12,800 --> 00:12:18,760 I do think that is reasonable. I would be really surprised if 225 00:12:19,120 --> 00:12:22,120 everybody's location was just naturally opted in. 226 00:12:22,120 --> 00:12:24,720 I don't know. But like if you guys have 227 00:12:24,720 --> 00:12:27,000 Instagram maps, can you please let me know? 228 00:12:27,120 --> 00:12:30,240 I think it was Macy from Secret Lives of Mormon Wives. 229 00:12:30,280 --> 00:12:33,720 She did a question box. Somebody said, hey, like your 230 00:12:33,720 --> 00:12:35,720 maps is on, we can see your location. 231 00:12:35,920 --> 00:12:39,680 She responded to it on her story saying I didn't turn it on, but 232 00:12:39,680 --> 00:12:43,160 also the location that meta map says that I am like I'm not 233 00:12:43,160 --> 00:12:45,960 there. So I don't understand how this 234 00:12:45,960 --> 00:12:48,080 works. If you guys have it, please let 235 00:12:48,080 --> 00:12:49,760 me know. Obviously there's some other 236 00:12:49,760 --> 00:12:52,360 issues with it. You know, apparently it updates 237 00:12:52,360 --> 00:12:55,120 your location whenever you're using the app, so that's 238 00:12:55,120 --> 00:12:56,520 something to have to think about. 239 00:12:56,800 --> 00:12:59,560 Second, it can be easy to share your location with users that 240 00:12:59,560 --> 00:13:01,920 you don't necessarily want to based on the settings. 241 00:13:02,160 --> 00:13:05,440 So Instagram users can share their location with friends and 242 00:13:05,440 --> 00:13:08,840 followers that you follow back. close friends only selected 243 00:13:08,840 --> 00:13:11,560 friends or no one. So obviously users should be 244 00:13:11,560 --> 00:13:13,720 careful you know who they select here. 245 00:13:14,040 --> 00:13:16,360 And 3rd. Some users allegedly saw their 246 00:13:16,360 --> 00:13:19,080 posts on the map before they opted in, according to the New 247 00:13:19,080 --> 00:13:21,640 York Times quote. One explanation offered by 248 00:13:21,640 --> 00:13:25,440 Instagrams leader Adam Masseri is that the map was populated 249 00:13:25,480 --> 00:13:29,560 not only with real time locations, but also with earlier 250 00:13:29,560 --> 00:13:32,560 posts tagged with a location. Either way, I feel like the 251 00:13:32,560 --> 00:13:37,680 comms around Instagram maps has been an actual nightmare and and 252 00:13:37,680 --> 00:13:40,960 they really should have been more considered about this 253 00:13:40,960 --> 00:13:44,920 rollout because people are so concerned about big tech 254 00:13:44,920 --> 00:13:48,120 surveillance about just in general privacy concerns right 255 00:13:48,120 --> 00:13:49,960 now. And I think rolling this out 256 00:13:49,960 --> 00:13:53,360 without very clear communications as to how it 257 00:13:53,360 --> 00:13:57,080 works, when it will update your location, who it can be updated 258 00:13:57,080 --> 00:13:59,480 for, if it's an automatic opt in or not. 259 00:13:59,480 --> 00:14:03,120 Like that needed to be like a widely spread message because 260 00:14:03,120 --> 00:14:05,520 there are so many like conspiracy theories going around 261 00:14:05,520 --> 00:14:08,080 about this. And of course, everybody hates 262 00:14:08,080 --> 00:14:10,360 Meta, so everybody is inclined to believe those. 263 00:14:11,120 --> 00:14:13,800 All right guys, that's it for mini scroll today, Mini tech 264 00:14:13,800 --> 00:14:15,520 scroll. Thank you so much for listening. 265 00:14:15,520 --> 00:14:17,720 I love you guys. Please remember to rate, review 266 00:14:17,720 --> 00:14:20,000 and subscribe and I will talk to you tomorrow. 267 00:14:20,320 --> 00:14:20,600 Bye.