1 00:00:01,800 --> 00:00:05,720 Speaker 1: A zone media hell, and welcome to this week's better 2 00:00:05,760 --> 00:00:20,080 Speaker 1: offline monologue. I'm your host ed Zetron. What last Sunday 3 00:00:20,160 --> 00:00:22,400 Speaker 1: was the Super Bowl, the final game of the NFL season, 4 00:00:22,440 --> 00:00:24,079 Speaker 1: And for those of you who don't like sport or 5 00:00:24,079 --> 00:00:26,319 Speaker 1: live in another country or something, it's an event that 6 00:00:26,360 --> 00:00:28,360 Speaker 1: many people just watch for the many, many ads that 7 00:00:28,360 --> 00:00:31,280 Speaker 1: they cut into during the game. The Super Bowl ad 8 00:00:31,320 --> 00:00:33,159 Speaker 1: can cost as much as eight million dollars for a 9 00:00:33,200 --> 00:00:35,760 Speaker 1: thirty second commercial, depending on where it air is during 10 00:00:35,800 --> 00:00:38,159 Speaker 1: the game, the most coveted slots being just before the 11 00:00:38,200 --> 00:00:41,360 Speaker 1: halftime show, and the ads themselves tend to be the spectacular, 12 00:00:41,400 --> 00:00:44,400 Speaker 1: highly produced fare. They're very weird, but they put a 13 00:00:44,400 --> 00:00:48,000 Speaker 1: ton of money into them. This year, I was specifically 14 00:00:48,040 --> 00:00:51,519 Speaker 1: excited to see what bullshit slot the various AI companies 15 00:00:51,560 --> 00:00:54,720 Speaker 1: would try and force down my nasty little throat. Last year, 16 00:00:54,800 --> 00:00:57,520 Speaker 1: anthropicaird a five second long commercial that said, and I 17 00:00:57,600 --> 00:01:01,720 Speaker 1: quote Claude is a next generation AI system, a sentence that, 18 00:01:01,840 --> 00:01:05,120 Speaker 1: when you think about it, means precisely nothing. And Microsoft 19 00:01:05,200 --> 00:01:07,920 Speaker 1: famously aired a co pilot AI commercial where several of 20 00:01:07,959 --> 00:01:10,120 Speaker 1: the prompts, such as write code for my three D 21 00:01:10,200 --> 00:01:13,440 Speaker 1: open World game simply didn't do what they said they did. 22 00:01:14,400 --> 00:01:16,040 Speaker 1: In the case of the Open World game by the way, 23 00:01:16,080 --> 00:01:18,720 Speaker 1: from last year, it generated a step by step guide 24 00:01:18,720 --> 00:01:21,080 Speaker 1: to building one, which isn't the same thing at all. 25 00:01:21,160 --> 00:01:23,360 Speaker 1: It's kind of like ordering dinner and Uber EAT's bringing 26 00:01:23,400 --> 00:01:26,160 Speaker 1: your recipe for a burrito instead of an actual burrito, 27 00:01:26,200 --> 00:01:30,000 Speaker 1: but still charging you all the same. Nevertheless, this year's 28 00:01:30,000 --> 00:01:32,720 Speaker 1: slot was even more obscene, starting with an ad for 29 00:01:32,760 --> 00:01:35,720 Speaker 1: Google's Pixel nine phone where a guy has asked questions 30 00:01:35,760 --> 00:01:38,320 Speaker 1: as if he's taking a job interview by Google's generative 31 00:01:38,319 --> 00:01:42,200 Speaker 1: AI Gemini Live. As he answers them, it's obvious that 32 00:01:42,319 --> 00:01:44,880 Speaker 1: what he's describing is actually being a father. They also 33 00:01:45,040 --> 00:01:47,360 Speaker 1: show you an overlay of like him taking his kid 34 00:01:47,360 --> 00:01:50,280 Speaker 1: to score. It's got emotional music playing with him taking 35 00:01:50,320 --> 00:01:53,200 Speaker 1: her to school, and at one point e Google's explain 36 00:01:53,360 --> 00:01:55,800 Speaker 1: bling to a child, which I should be clear is 37 00:01:55,840 --> 00:01:59,600 Speaker 1: not a Gemini product, it's just Google Search. It's unclear 38 00:01:59,680 --> 00:02:02,720 Speaker 1: what this ad is actually advertising. We don't hear the 39 00:02:02,760 --> 00:02:05,240 Speaker 1: prompt that's started these questions, nor do we know why 40 00:02:05,280 --> 00:02:07,880 Speaker 1: he's answering them, but when he's done, he's clearly joining 41 00:02:07,920 --> 00:02:11,640 Speaker 1: some sort of job interview. Virtually Gemini doesn't really respond 42 00:02:11,639 --> 00:02:13,680 Speaker 1: to any of his answers other than to say things like, 43 00:02:13,720 --> 00:02:16,400 Speaker 1: try rephrasing your answer to sound more confident, so we 44 00:02:16,440 --> 00:02:19,119 Speaker 1: can assume. And that's all we're doing here that he's 45 00:02:19,160 --> 00:02:22,520 Speaker 1: preparing for a job interview. He's also pretty old, like 46 00:02:22,600 --> 00:02:24,399 Speaker 1: his kid has gone off the college. She's a full 47 00:02:24,400 --> 00:02:28,959 Speaker 1: grown adult. Now. Very weird, very weird commercial, very clearly 48 00:02:29,000 --> 00:02:32,200 Speaker 1: built for people who cry whenever they remember their parents. Sorry, 49 00:02:32,200 --> 00:02:35,839 Speaker 1: if that's you you are, do not use Gemini Life. 50 00:02:35,960 --> 00:02:40,160 Speaker 1: That's my recommendation. Next up was open a Eyes gratuitously 51 00:02:40,240 --> 00:02:44,320 Speaker 1: stupid commercial called The Intelligence Age, which showed a dot 52 00:02:44,400 --> 00:02:49,000 Speaker 1: matrix history of time involving men throwing spears, people riding horses, 53 00:02:49,040 --> 00:02:52,000 Speaker 1: waves crashing a waving skeleton for some reason as a plane, 54 00:02:52,000 --> 00:02:54,360 Speaker 1: and eventually a globe with a few dots going toward 55 00:02:54,360 --> 00:02:57,640 Speaker 1: a computer with a modem sound playing. A full forty 56 00:02:57,800 --> 00:03:01,400 Speaker 1: seconds in to this one minute long commercial, we start 57 00:03:01,440 --> 00:03:04,840 Speaker 1: actually hearing prompts. Someone says, summarize this article for me, 58 00:03:04,960 --> 00:03:08,040 Speaker 1: Then someone else says, help me practice asking for a raise. 59 00:03:08,400 --> 00:03:10,720 Speaker 1: And this prompt, by the way, is near impossible to hear, 60 00:03:10,919 --> 00:03:12,839 Speaker 1: as it's cut off to have a guy say act 61 00:03:12,880 --> 00:03:15,840 Speaker 1: as my sleep coach. And then there's two more prompts 62 00:03:15,880 --> 00:03:18,840 Speaker 1: in a language I don't speak, then another that says 63 00:03:18,880 --> 00:03:21,200 Speaker 1: turn my idea into a business plan, before a chirpy 64 00:03:21,320 --> 00:03:24,640 Speaker 1: voice asks what do you want to create next? This 65 00:03:24,800 --> 00:03:27,280 Speaker 1: is really good. I love this. I also love the 66 00:03:27,320 --> 00:03:30,760 Speaker 1: idea of having two prompts in different languages, because people 67 00:03:30,800 --> 00:03:33,760 Speaker 1: speak different languages other than English. I understand that there 68 00:03:33,800 --> 00:03:38,760 Speaker 1: is no official language of America, so it's good. But nevertheless, 69 00:03:38,800 --> 00:03:42,240 Speaker 1: that you only got a few seconds. You didn't think 70 00:03:42,320 --> 00:03:44,280 Speaker 1: to try and get as many prompts in. Maybe it's 71 00:03:44,320 --> 00:03:46,400 Speaker 1: that they didn't have them. In fact, if you're a 72 00:03:46,480 --> 00:03:50,080 Speaker 1: listener and you watch this commercial and you speak those languages, 73 00:03:50,120 --> 00:03:52,920 Speaker 1: please email me. It's easy at better offline dot com. 74 00:03:52,920 --> 00:03:55,720 Speaker 1: I would love to hear what they actually said. Nevertheless, 75 00:03:56,040 --> 00:03:58,880 Speaker 1: asking chat GPT to be your sleep coach leads it 76 00:03:58,920 --> 00:04:01,920 Speaker 1: to ask you about your sleep habits, your lifestyle, common issue, 77 00:04:01,960 --> 00:04:05,200 Speaker 1: sleep quality, and your ideal goals. Asking it to turn 78 00:04:05,240 --> 00:04:07,680 Speaker 1: my idea into a business plan immediately led it to 79 00:04:07,720 --> 00:04:11,600 Speaker 1: asking me what my idea was. Great. I can practice 80 00:04:11,600 --> 00:04:14,040 Speaker 1: with a large language model how to ask for a raise, 81 00:04:14,400 --> 00:04:17,440 Speaker 1: But what does that mean exactly? What does this product, 82 00:04:17,480 --> 00:04:20,400 Speaker 1: do what is chat GPT and assuming that AI ends 83 00:04:20,480 --> 00:04:22,919 Speaker 1: up actually wrecking the labor market as they wanted to, 84 00:04:23,240 --> 00:04:25,840 Speaker 1: the idea of using chat GPT is a soundboard when 85 00:04:25,880 --> 00:04:28,400 Speaker 1: you renegotiate your contract kind of feels like more of 86 00:04:28,440 --> 00:04:31,120 Speaker 1: a cruel joke. It's also not going to do that. 87 00:04:31,160 --> 00:04:34,440 Speaker 1: But anyway, two fucking years into this nonsense and these 88 00:04:34,440 --> 00:04:37,599 Speaker 1: assholes can't even come up with one interesting or useful example. 89 00:04:37,640 --> 00:04:40,479 Speaker 1: It's a fucking joke. These people have such contempt for 90 00:04:40,600 --> 00:04:43,080 Speaker 1: customers that they can barely be bothered to advertise their 91 00:04:43,080 --> 00:04:47,520 Speaker 1: products also goddamn half hearted. Yet the worst one by 92 00:04:47,680 --> 00:04:52,320 Speaker 1: far was Salesforce and their Gait Expectations ad, where Matthew McConaughey, 93 00:04:52,360 --> 00:04:54,719 Speaker 1: an actor, says that he and I quote didn't use 94 00:04:54,720 --> 00:04:58,080 Speaker 1: Agent Force, the powerful AI from Salesforce, so an AI 95 00:04:58,200 --> 00:05:01,039 Speaker 1: agent didn't send him the fastest to his gait or 96 00:05:01,120 --> 00:05:04,160 Speaker 1: root of your British which has now changed. The commercial 97 00:05:04,279 --> 00:05:07,160 Speaker 1: ends by saying that Salesforce helps London Heathrow create a 98 00:05:07,279 --> 00:05:12,599 Speaker 1: first class experience. Everyone in the commercial was also American act. Nevertheless, 99 00:05:12,760 --> 00:05:15,080 Speaker 1: I n actually went and looked up what this integration 100 00:05:15,240 --> 00:05:19,440 Speaker 1: actually looked like and discovered that Heathrow actually integrated Einstein's chatbots, 101 00:05:19,720 --> 00:05:23,240 Speaker 1: Einstein being Salesforce's large language model chatbot thing. They've been 102 00:05:23,440 --> 00:05:26,760 Speaker 1: warming up for that a decade, but this was back 103 00:05:26,760 --> 00:05:30,719 Speaker 1: in May twenty twenty three. They claimed that Einstein answered 104 00:05:30,760 --> 00:05:33,200 Speaker 1: more than four thousand questions a month and helped cut 105 00:05:33,240 --> 00:05:36,000 Speaker 1: call service volumes by twenty seven percent, and that Heathrow 106 00:05:36,040 --> 00:05:38,240 Speaker 1: has seen a four hundred and fifty percent increase in 107 00:05:38,320 --> 00:05:42,240 Speaker 1: live chat usage since launch. Then they proudly added that today, 108 00:05:42,320 --> 00:05:45,400 Speaker 1: and the article is completely undated, Heathrow had seen about 109 00:05:45,800 --> 00:05:49,479 Speaker 1: around forty to sixty second quicker per contact interactions in 110 00:05:49,480 --> 00:05:53,920 Speaker 1: call centers with Einstein chatbots. This is the only data 111 00:05:54,000 --> 00:05:57,159 Speaker 1: that appears to exist for anything to do with Agent 112 00:05:57,279 --> 00:06:00,440 Speaker 1: Force and Heathrow. There's nothing else they could find. I 113 00:06:00,480 --> 00:06:03,480 Speaker 1: will add that a record breaking eighty three point nine 114 00:06:03,600 --> 00:06:07,680 Speaker 1: million passengers flew through Heathrow in twenty twenty four four 115 00:06:07,720 --> 00:06:10,280 Speaker 1: thousand questions a month or forty eight thousand a year. 116 00:06:10,680 --> 00:06:12,240 Speaker 1: It's kind of a drop in the buck in as 117 00:06:12,240 --> 00:06:15,679 Speaker 1: a percentage at zero point zero five seven two percent, 118 00:06:16,880 --> 00:06:20,120 Speaker 1: And like I said, really not finding any information about 119 00:06:20,160 --> 00:06:23,000 Speaker 1: their Agent Force integration. All I could find was a 120 00:06:23,040 --> 00:06:25,400 Speaker 1: LinkedIn post from Sharon Pryor, who is the director of 121 00:06:25,480 --> 00:06:28,080 Speaker 1: Technology at Heathrow, who said that four hundred percent more 122 00:06:28,160 --> 00:06:31,520 Speaker 1: questions were handled via live chat, which is interesting because 123 00:06:31,520 --> 00:06:34,080 Speaker 1: the one from twenty twenty three it seems four hundred 124 00:06:34,080 --> 00:06:36,640 Speaker 1: and fifty percent. I guess that went down. And apparently 125 00:06:36,680 --> 00:06:40,479 Speaker 1: Agent Force says something called instant FAQs and interminal wayfinding 126 00:06:40,560 --> 00:06:43,680 Speaker 1: for your gate plus restaurant shops and facilities. I want 127 00:06:43,720 --> 00:06:45,480 Speaker 1: to be clear that all this bullshit was doable with 128 00:06:45,560 --> 00:06:47,800 Speaker 1: knowledge management systems and chatbots all the way back in 129 00:06:47,839 --> 00:06:51,279 Speaker 1: the mid twenty tens. Salesforce isn't even describing a AI agents, 130 00:06:51,279 --> 00:06:55,360 Speaker 1: which are supposedly autonomous chatbots. They're describing this boring, bob 131 00:06:55,600 --> 00:06:58,279 Speaker 1: standard bullshit. We can help you find your gate, stuff 132 00:06:58,320 --> 00:07:01,320 Speaker 1: that apps have done for a decade. What's more insulting 133 00:07:01,400 --> 00:07:05,400 Speaker 1: is the commercial officscates what this product actually is. Macarache 134 00:07:05,600 --> 00:07:08,360 Speaker 1: as a traveler is saying he didn't use Agent Force, 135 00:07:08,400 --> 00:07:10,840 Speaker 1: which is the powerful AI from Salesforce, as if that 136 00:07:10,920 --> 00:07:13,840 Speaker 1: is something a consumer chooses to do versus a business 137 00:07:13,840 --> 00:07:16,720 Speaker 1: being actively conned by Mark Benioff, the CEO of Salesforce. 138 00:07:17,360 --> 00:07:20,120 Speaker 1: Salesforce has no powerful AI. It's all a lie, a 139 00:07:20,160 --> 00:07:22,280 Speaker 1: con an attempt to pretend that a company does something 140 00:07:22,320 --> 00:07:25,320 Speaker 1: that they do not do. Salesforce has a large language 141 00:07:25,320 --> 00:07:28,400 Speaker 1: bottel they're working on. They've used chat GBT or GBT 142 00:07:28,480 --> 00:07:30,920 Speaker 1: four in the past. Nevertheless, I don't know what the 143 00:07:30,960 --> 00:07:35,040 Speaker 1: powerful AI is. Salesforce, well, their problem is simple. They 144 00:07:35,040 --> 00:07:36,920 Speaker 1: can't do an AD that says, hey, do you want 145 00:07:36,920 --> 00:07:39,440 Speaker 1: a slightly more powerful chatbot that costs a great deal 146 00:07:39,480 --> 00:07:41,200 Speaker 1: of money, So they have to make this kind of 147 00:07:41,200 --> 00:07:44,080 Speaker 1: incongruent AD that lies about their product without ever really 148 00:07:44,120 --> 00:07:47,960 Speaker 1: saying what it does. Open AI, two years into this bubble, 149 00:07:48,080 --> 00:07:51,160 Speaker 1: still can't muster up a single meaningful, interesting use case. 150 00:07:51,320 --> 00:07:54,400 Speaker 1: And in both Google and AI's case, it truly isn't 151 00:07:54,440 --> 00:07:57,280 Speaker 1: obvious what the product is that they're selling. And that 152 00:07:57,400 --> 00:08:00,880 Speaker 1: really is the ultimate summary of the AI boos. Vague 153 00:08:00,920 --> 00:08:03,600 Speaker 1: and blatant lies that barely masked the contempt that these 154 00:08:03,600 --> 00:08:06,560 Speaker 1: companies have for customers, so much content that they can't 155 00:08:06,600 --> 00:08:09,440 Speaker 1: even bother to give them a real reason to use it, 156 00:08:09,920 --> 00:08:12,280 Speaker 1: And if a monest I'm a little bit disappointed about 157 00:08:12,320 --> 00:08:14,600 Speaker 1: these commercials because they lacked even the juice of the 158 00:08:14,640 --> 00:08:17,800 Speaker 1: twenty twenty two Crypto Bacanalia. You see there, we had 159 00:08:17,840 --> 00:08:20,080 Speaker 1: celebrities that called you a pig and a moron. They 160 00:08:20,080 --> 00:08:21,920 Speaker 1: said you were a hog that rolled around in the 161 00:08:21,960 --> 00:08:25,160 Speaker 1: mud and you oinked and squealed instead of putting everything 162 00:08:25,480 --> 00:08:28,520 Speaker 1: you had behind Crypto and people like Tom Brady while 163 00:08:28,520 --> 00:08:30,960 Speaker 1: they delivered, or at least on a superficial level, these 164 00:08:31,000 --> 00:08:35,200 Speaker 1: compelling ish messages that suggested you might really need to 165 00:08:35,240 --> 00:08:38,720 Speaker 1: and successfully convince people to buy these nebulous tokens in NFTs, 166 00:08:39,000 --> 00:08:41,240 Speaker 1: even though many of the companies they promoted turned out 167 00:08:41,240 --> 00:08:45,080 Speaker 1: to be elaborate Ponzi schemes such as FDx. And never 168 00:08:45,120 --> 00:08:47,640 Speaker 1: have I been more certain that this bubble will pop. 169 00:08:48,480 --> 00:08:51,880 Speaker 1: This was the biggest stage that generative AI will ever have, 170 00:08:52,200 --> 00:08:55,120 Speaker 1: and they blew it because well, deep down, they realized 171 00:08:55,559 --> 00:08:56,920 Speaker 1: there's just not that much to sell.