The Human Use of Human Beings

Brian: [00:00:56] Hello and welcome to Future Commerce, the podcast about the next generation of commerce. I'm Brian.

Phillip: [00:01:04] I'm Phillip, and we got some breaking news. But before we do, hey, I'm just going to bring over the tone from our Senses emails that we've been sending out here since the beginning of the year. I set the tone. I didn't ask you permission, by the way. I should have probably done that. I wanted to set the tone for the beginning of the year, Brian, as relentless optimism was going to be the drum we're banging for 2023. You can't keep a good dog down. I feel like CES was pretty optimistic. The NRF Big Show was pretty optimistic. And with that in mind, wow, there's a lot of layoffs happening right now in a lot of places.

Brian: [00:01:47] Oh, we sit here laughing about layoffs, but it's only in the context of us being optimistic. I'm still optimistic about the year. There are some weird things about these layoffs. I mean, the fact that every major tech company is laying people off except for Apple. I mean, you can't help but think that there's a little bit of an excuse scenario going on here. Oh, everyone's laying off. Now's our time to get rid of people." It's very hard to fire people. [00:02:20] I can't help but think that there's a lot of low-performer culling that's happening right now. However, that does not feel true for the Google layoff. To me, that was kind of an outlier in all of this because my understanding and this is both what I've read about, but also anecdotal, is that they let go a lot of high performers, long-time employees, and people who have been recently promoted. That is weird. [00:02:52]

Phillip: [00:02:52]  [00:02:52]I don't think it's weird. I'll tell you what I think it is. I think it's an existential shift of focus. There are two things that we have to kind of keep in mind. One is this thing I've been calling, which you heard in the last episode, Crossing the Rubicon, which is people just I think there's apathy around the things that we're used to. And people are tired of things and people have just decided that it's Google's time to be disrupted, and they're looking for reasons that Google should be disrupted, not because of a lack of quality in the service, not because the world can't find the information at the world's fingertips, which was always Google's mission. Google has organized the world's information. They have done that. They did what they set out to accomplish. I think there is a certain kind of person in our industry who is ready for the challenger. And last year they were beating the drum, saying TikTok Search, especially among Gen Z, will displace Google Search for the next generation. People will go to TikTok first, which by the way, I have a whole thesis on that. I feel like I haven't gone in-depth on the podcast. [00:04:10]

Brian: [00:04:10]  [00:04:10]Everybody wants Google to die. That's what it sounds like. Everyone's like... [00:04:15]

Phillip: [00:04:16]  [00:04:16]They're calling for its head. [00:04:16]

Brian: [00:04:16]  [00:04:16]"This is the Google killer." Everybody just kind of wants them to lose. [00:04:19]

Phillip: [00:04:19]  [00:04:19]Everything's a Google killer. [00:04:20]

Brian: [00:04:20]  [00:04:20]Yeah. {laughter}

Phillip: [00:04:21]  [00:04:21]But in reality and they are it's because it's a David and Goliath scenario. You know, Google, won de facto Google and Apple sort of own a giant part of tech. So Google is sort of retrenching. There are conversations that Sundar Pichai is personally heading the Google Duplex initiative, which is their ChatGPT killer, theoretically.  There's a whole large language model which has been being developed for years now inside of Google, which they have yet to pull back the curtain on, so much so that we've reported often here on the show. But if you're just listening for the first time or you're catching up, Blake Lemoine, a researcher at Google, went to a US senator's office late last year, in 2022, and claimed that this language model they had developed had become sentient and was self-aware and that they have created an AI that is terrifying and powerful and it should be stopped at all costs. This is theoretically Google's answer to the OpenAI initiative, which Microsoft has aligned itself around. So, yeah, there are battle lines being drawn. People want to take down the giant. So it doesn't feel all that inexplicable to me, Brian. Layoffs are refocusing where the priorities are in one part of the business to futureproof itself for the next generation. That's how I would explain it. [00:05:46]

Brian: [00:05:46] Yeah, that makes sense.

Phillip: [00:05:47] It's not going to come through live streaming or merch sales on YouTube because that's not where the levers are. I think that's where a lot of the commerce analysts industrial complex wanted the growth to come from. But that's not where the growth is going to come from.

Brian: [00:06:02] I mean, I think that long term you're absolutely right. There's monetization to be had for existing products, and Google's done a great job of that over the years. Let's be clear, Google...

Phillip: [00:06:13] Value extraction.

Brian: [00:06:15] {laughter} And I think they're going to probably continue to do that. But I think that they're looking ahead in they're like, "Oh, wow, actually this may be the actual Google killer. And our competitor, Microsoft, just invested a lot of money in it." And I think Microsoft, probably a similar scenario where they laid off 10,000, another big layoff there. They're refocusing. They did that the same week they announced like $100 Million investment in OpenAI. Refocusing is probably the term. The one thing that kind of caught me off guard again about the Google pieces, though, is I've seen other big tech companies, tech companies like Adobe and others, they usually do a really good job of moving high-value individuals. Individuals that they see are really talented, people that have just been promoted or that have added a lot of value in other places. They move them around in the company. And so that was the Google one... I do think there might have been some actual cost-cutting happening. There were reports of people who were earning $1,000,000 annually that got cut.

Phillip: [00:07:28] I'm sure they'll be okay.

Brian: [00:07:30] I'm sure they'll be okay, too. That's not my point. My point is saying, I actually think Google was going to town when it came to...

Phillip: [00:07:38] Oh yeah. Well, let's not forget that I think there was a report in the Wall Street Journal that Google had almost doubled its workforce since the pandemic. This is just a... And it's like any cuts that are being made in tech right now are just slightly pulling back from the explosion and workforce additions that they have made over the last two and a half, three years. So I wouldn't be sounding the alarm. Do you know what we also have at the same time is a historically low unemployment rate in the United States of America. So and that's even notwithstanding the fact that nobody wants to work in fast food anymore, it's all going to be machines and automation. I have to go to a kiosk. I can't find the hamburger with no cheese because my kid hates cheese. I have to stand at the kiosk for five, six, 7 minutes staring at a menu screen of options. Not like it is markedly a worse customer experience, but that's the world that we all chose. So this is the world we all have to live in now. And we still have historically low unemployment. So I don't know.

Brian: [00:08:39] By the way, I talked to someone that went to the McDonald's that's fully automated and he said it was terrible because...

Phillip: [00:08:44] I love it.

Brian: [00:08:45] There wasn't even an attendant there.

Phillip: [00:08:46] Give us fewer reasons in this world to go to McDonald's. I don't have any problem with that.

Brian: [00:08:52] I think you're dead on. Another really good point here is that these companies have taken some serious knocks on the chin from the street this past year. The stock market loves layoffs.

Phillip: [00:09:03] Oh, yeah, no. And it's all factored in. This is all you know, it's a lot of consternation over what I think the All In Podcast would call surplus elites. We got a lot of people making a lot more money than they probably deserve. I was probably one of them at one point in time. And now I run a podcast and that's okay. But I think let's come back to Commerce because I think this is a really important point to make. Why does all of this matter? I think we can tie it up. [00:09:34] What really matters is a lot of the things that have been repeated as the next generation-defining commerce experiences are just going to flat out turn out to be wrong or untrue or wishful thinking. And in particular, it's what were the major growth levers that we were going to anticipate in the next ten years? Let's look at what the analyst industrial complex said. If you look at McKinsey and you look at Gartner and you look at Forrester and you look at Coresight and you look at what other analysts have predicted around eCommerce growth and retail growth, they would tell you it's experiential, it's metaverse, it's AR, it's VR, it is cryptocurrency, it's Web3 experiences. And it turns out that maybe none of that's true. They would tell you it's Far East shopping experiences, it's conversational commerce, text message based, it's super apps, it's live stream shopping, it's influencer... And while all of those have some facet of truth to them in that they influence the way that we want to engage... They tell you it's social commerce and you're going to buy stuff direct through Instagram and that's going to happen. Tiktok's going to launch fulfillment centers in the United States. And all of this is subject to well, yeah, if that's what the consumer wants. But it turns out that's not what the consumer wants. The consumer wants none of those things. On no one's radar was a conversational chatbot that writes your job application for you. And I think that this democratization and consumer access of AI and language generation is going to have a material impact on the way that we buy things [00:11:16] because we are already seeing... We had a conversation, Brian, at a Future Commerce salon that we had in New York last week with a CEO of a major fashion house, saying that within two weeks of getting access to GPT3, they were using it in production. I've never seen the adoption of technology, and albeit freemium technology, that has made its way into in front of consumers at such a clip. And not a single analyst could have told you that in the last few years because everyone was bought into a hype cycle. By the way, one that I have said was patently false for years now.

Brian: [00:11:54] You have. For years you've been beating that drum.

Phillip: [00:11:55] I'm on record saying social commerce is a farce.

Brian: [00:11:59] Livestream shopping is dead.

Phillip: [00:12:00] Livestream shopping is a farce. It's not going to happen. And I'm okay with being definitive and very clear that these are wish fulfillment. Just because something's happening in one part of the world doesn't mean it's destined to come to another. It really depends on how the American consumer, at least in our perspective, because that's where we're based and that's the primary audience we serve. It's like how they adopt technology and those are much longer adoption cycles than I think anybody is prepared to accept because it doesn't make for like a really sexy annual report and trends report. That's why.

Brian: [00:12:34] Exactly. Well, what's actually happening is now finally spurring me to read the book that I've been wanting to read for a while, which is...

Phillip: [00:12:46] Archetypes.

Brian: [00:12:46] {laughter} Archetypes.

Phillip: [00:12:46] ArchetypesJournal.com. That's the one you've been... No.

Brian: [00:12:49] I mean, I don't have to read that one. I wrote it with you. But I actually I have read Archetypes over and over now. It's so good.

Phillip: [00:13:02] What's the actual book you were going to reference? Sorry for the shameless plug.

Brian: [00:13:04] The Human Use of Human Beings because ChatGPT is crazy. I think there's a lot of discussion right now about what our role in work is going to be. And there's actually a lot of talk in The Human Use of Human Beings and some other of Norbert Weiner's books that is focused on automation and displacement of jobs, actually, and also where humans and machines need to interact and where humans and humans can interact and where machines to machines need to interact. And so there's a lot of thought around, and I'm still in the middle of the book. I'm only like 20 pages in right now, so I'll have to give a full...

Phillip: [00:13:55] It's always the best way. To give a full book review.

Brian: [00:13:59] Exactly. I'm actively learning about the role that we played in and actually Wiener heavily influenced Marshall McLuhan, who we've referenced pretty heavily lately.

Phillip: [00:14:41] I have to make a public apology. I want to give a plenty of space so that everybody can let that sink in for a second. I'm realizing now and it took six years to do it, so I apologize. You're like way out in front on a bunch of stuff. So much so that I'm like, I have no idea what you're talking about. Stop talking about it. But it turns out that after about six months of marinating in the bigger ideas, I'm like, I finally kind of come around to things. You're on to some really wild things. I have to turn our audience on before we come back to The Human Use of Humans. Is that the title?

Brian: [00:15:17] The Human Use of Human Beings, yeah.

Phillip: [00:15:19] Human beings. Okay. I want to come back to that because I have a lot to say. But you wrote a piece called Quantum Yeet, which... Still don't love the title.

Brian: [00:15:31] I said we should change it. Everyone was like, "Oh, it's too late, it's too good."

Phillip: [00:15:35] It's fine.

Brian: [00:15:35] And I was like, "I don't think this title is good."

Phillip: [00:15:37] It gets attention. I sat in a meeting last week with one of our listeners and a reader of ours of the newsletter who said his favorite piece from the last couple of months was Quantum Yeet. And I almost fell out of my chair because I told you at the time, I just don't get this dude. And I'm a person who reads quantum mechanics books for fun. But I just didn't understand the points that you were making, and it took me a little while, but he explained it back to me in a way that I felt like was really beneficial for me to understand. And it also kind of showed me that it's really hard to be your audience surrogate when you're a content creator. It's like really hard to put yourself in the position of the audience because I just wasn't jamming on it. But he had this really great way of explaining [00:16:28] the customer journey is not linear, [00:16:30] and if you just use that phrase right there, this is by the way, this is like if you're not onto this stuff in the Future Commerce oeuvre, you should be because the big, big, big ideas usually make their way into our larger pieces. And we're thinking pretty heavily right now about our Visions report this year and what that looks like. So it's all influencing, right? But he said the customer journey is not linear from the inspiration to purchase something to the actual fulfillment of the purchase and then the gratification you receive from it. You are actually traversing along many retailers timelines. So if Amazon were to interrogate the purchase that you're looking to make, they see a discrete start and stop. But that is not the true start and stop, right? They see you're shopping for it and your purchase of it, but that wasn't where the original inspiration was. And part of our jobs as marketers is to try to make sense of this journey and to attribute that journey. But you probably shopped maybe at Target for the same thing or maybe on Etsy and maybe you ask some friends about it. In reality, you existed across a multitude of marketing funnels at one time. It is quantum superpositions and it all has to do with the point of reference. So without using the word quantum, a given marketer from their point of reference has one perspective of your journey. If you were to put yourself in God mode to see the entirety of the journey, you'll see that you're actually moving between multiple timelines. That is quantum. And when Kyle re-explained this to me, I'm like, "I finally understand it." Why doesn't Kyle write that piece, Brian?

Brian: [00:18:09] I don't know. Let's have Kyle... You should have Kyle write for us.

Phillip: [00:18:12] Let's have Kyle write for us from now on.

Brian: [00:18:14] I don't even know what I'm talking about half the time.

Phillip: [00:18:16] I love it. Well, I think it's really freaking brilliant. And and so rather than make fun of you about Marshall McLuhan, I won't do it again. Rather than beat you over the head with the word quantum, I'm just going to apologize and say...

Brian: [00:18:31] Oh sheesh.

Phillip: [00:18:32] I haven't gotten it. But now I think I get it.

Brian: [00:18:35] The thing is, I think the world of McLuhan is a lot deeper than I have even touched. And actually the world of quantum is going to be also bigger than... These are words that have worked their way into... Quantum is something that's a concept that's worked its way into our concepts as a society. Quantum Mania. The third Ant Man is coming out soon.

Phillip: [00:19:03] Oh gosh. That's happening now.

Brian: [00:19:03] But I think that they're actually reflective of bigger ideas that are working their way into pop culture, that are preparing us for how we have to approach the world coming up.

Phillip: [00:19:15] They're desensitizing us to the coming zombie apocalypse that we're all about to encounter.

Brian: [00:19:22] {laughter} I don't know about zombie apocalypse, although watching The Last of Us, that seems likely now.

Phillip: [00:19:28] So I want to put a bow on this because The Human Use of Human Beings... You touched on a subject about the displacement of jobs, in particular, probably marketing jobs and maybe in particular content creation and creative roles. And it's funny because we just mentioned the McDonald's kiosk. The McDonald's kiosk displaces a job in the economy that arguably very few people wanted to begin with.

Brian: [00:19:59] Correct.

Phillip: [00:20:00] They were hard to staff. They were hard to retain. Almost nobody wanted to work in that role. And those I mean, entry level roles in an economy are necessary and vital. And I don't want to disparage them. But in a hundred year old business like McDonald's, at some point, very few people want that job to begin with. I have to believe that very few people want the job of SEO Writer. No one wants to write a 6000 word recipe blog that is laden with a bunch of SEO garbage and jargon. You don't really want that job. Is it a job? Yes. But nobody really wants that job. They have to do that job. So this is where I think we're going. The McDonald's kiosk was probably the labor of hundreds of industrial designers who had to create electronics and touchable interfaces and probably dozens of interface designers and graphic artists and integration specialists and agencies and installers. Think about the hundreds and thousands of people and labor that went into creating and maintaining and updating. And it's more knowledge and information work to create a McDonald's kiosk. This is what's going to happen for the SEO blogger is that the new role is prompt engineer. We have people who understand the inputs and outputs and can write prompts for your SEO copywriting engine powered by GPT. And maybe not just even SEO. Maybe it's pricing and merchandizing strategies. Maybe it's your promotions calendar. Maybe it's your follow up emails and your abandoned cart campaigns and your welcome series. And these are all things that are sort of rote rinse and repeat playbooks that get implemented over and over that is not exactly engaging copy or even intellectually stimulating work that everybody needs, but almost nobody really wants to do on a daily basis.

Brian: [00:22:02] This is exactly right. In fact, I would argue that remember back when Google used to be like this weird mix of black and gray hat, just the weirdest, the weirdest industry sort of like opened up as a result of of Google search marketing the business.

Phillip: [00:22:20] Yes.

Brian: [00:22:21] That was actually people trying to figure out and satisfy machines. Algorithms. It was people's exploration of all the possible ways.

Phillip: [00:22:34] Exploitation.

Brian: [00:22:36] Exploitation.

Phillip: [00:22:37] Not exploration. I mean, it is exploration. It's exploitation of algorithms.

Brian: [00:22:41] Right. But they were trying to like basically, like hit every possible way that the algorithm could absorb information and figure it out, like figure out where they could best hit at that algorithm. Now, the thing is, the algorithm's intent by its creators wasn't necessarily to have an algorithm that people were just going to kind of blast around. They were trying to get an outcome out of it.

Phillip: [00:23:08] To make you cry on LinkedIn.

Brian: [00:23:10] Yes.

Phillip: [00:23:10] So that you get a number of likes and reshares.

Brian: [00:23:13] Bingo. The thing is, this was always best suited to be accomplished by a [00:23:18] machine. We as humans have been trying to accomplish jobs that are better done by machines. I just mentioned machine to machine communication. All the jobs that are best done by machines should be pursued as jobs that are done by machines. [00:23:37]

Phillip: [00:23:40] You remember we've referenced it a couple of times now, but there was an episode with Brian Roemmele. Hundreds of episodes. I mean, it's many years ago on the podcast.

Brian: [00:23:51] This is 19 and 20.

Phillip: [00:23:53] Episode 19, Episode 20. Yeah, there was a... What was he talking about? He had like voice computers that were all talking to each other in a closed room?

Brian: [00:24:03] Yeah. He had written a series of prompts that would spur additional prompts between I think it was Google and Alexa, or maybe it was just Alexa and Alexa, and he said that he had them in a closet, like barricaded, just prompting, prompting and prompting.

Phillip: [00:24:24] And they were having a conversation between each other.

Brian: [00:24:26] Right.

Phillip: [00:24:27] {laughter} Is what he claimed. And I thought it was far fetched, but that was five years ago, maybe six. And it's not so farfetched anymore. I just looked for it on the site. I can't find it. It's probably there. It's Episode 19, Episode 20. I hope you have three and a half hours to dedicate your life to those two episodes because it's a long, it's a long haul.

Brian: [00:24:48] Worth the listen though.

Phillip: [00:24:48] He outlined, if there's anyone that's a futurist that we know, like a true futurist, it's Roemmele. We should get him back on the show.

Brian: [00:24:58] We really should. I think we've said this a couple of times.

Phillip: [00:25:01] We need like a Tim Ferriss style, like five hour block for that guy. That's what we need.

Brian: [00:25:06] We should just create a new property just for Roemmele.

Phillip: [00:25:08] {laughter} Just...

Brian: [00:25:09] We do it all in a week.

Phillip: [00:25:10] It's a perpetual 90 hour exploration.

Brian: [00:25:13] We should have him interact with ChatGPT and then publish that as the interview.

Phillip: [00:25:20] I would watch it. I would watch it. I watch people cut sand on YouTube, for crying out loud. I'd listen to Brian Roemmele chat with a computer.

Brian: [00:25:29] You do a lot of ChatGPT.

Phillip: [00:25:31] I do.

Brian: [00:25:31] You're very good at it. You're very good at leading the witness as you will. And that's how you make it work.

Phillip: [00:25:39] So actually, I've been on to this for a little while. I want to kind of paint the picture of the future that I think that we're in because this human use of human beings, displacement of jobs, I heard this already once two decades ago when I got into eCommerce. The promise of eCommerce, I've said to so many times on this show, but [00:25:59] the promise of eCommerce was that we were going to have cheaper goods available faster because we are removing the unnecessary parts of labor and infrastructure in the retail purchase process, which means it's a better outcome for the consumer. So win/win. We are removing middlemen. We were going to take away the air conditioned showroom because who needs that? Well, that's only open a certain number of hours and it costs a lot of money to staff. And then you have to change the window dressing because people get bored. And you had all of these encumbrances to growing a retail footprint, retail business that included heat and gas and cooling and staffing. It was like none of that's necessary anymore because you can just shop 24/7 online Amazon.com. Everything you want for cheaper. But it turns out that's not the case because eCommerce is nothing but middlemen. It's laden with middlemen. In fact, 90% of the interactions on Twitter that I see are how many middlemen can you sign up for on your Shopify site today? It's the 16 apps you need to run your business. It's how many middlemen can you pack between yourself and the customer that can all become margin, extractive and reduce your profitability so that you can keep up with whatever fad or fashion is at the time? [00:27:15] I have heard this once before, Brian Lange. I feel like that's where we're heading again, because, yes, we are going to remove, we're removing the copywriter and the SEO analyst and we are removing some of the CX people. But what we are going to add as a result are prompt engineers, people that write briefs, people that write exacting detail about the website visuals that need to come out the other end and you're just moving one job from one place in the economy that has become so competitive that wages have gone down and the numbers of people that fulfill those roles and the expectations and the frequency that we see those roles and the availability of those roles has multiplied and we're going to get rid of those, and we're going to move them into prompt engineers and machine vision specialists and people who are really good at certain visual styles and coaxing that out of mid-journey.

Brian: [00:28:06] Coaxing being the right word.

Phillip: [00:28:08] And all we're doing is we're creating a new set of skills that has the same effect in output but is yet more middleman.

Brian: [00:28:15] It's not the same effect in output. So that's the one place where I think I would disagree. The outputs actually going to be better because we're using machines to accomplish things that we are not very good at. It took us 20 years to figure out how to get the SEO algorithm right, and that was due to thousands upon thousands of people poking at that algorithm for years and years and years. Those jobs should be done by machines. And so I think the outcome, what you're saying is we're not losing jobs. There's going to be jobs.

Phillip: [00:28:48] Where are you going to manage all these prompts?

Brian: [00:28:50] Right.

Phillip: [00:28:50] You're going to have them in a prompt manager. Like we need a PRM. That's what we need. The enterprise is going to have a prompt relationship management tool and you're going to store all your prompts in there and that's going to go into your machine-assisted output visual management tools. Like all of your brand visuals are going to go under there. It doesn't fit in a PIM and doesn't fit in a CRM, it doesn't really fit into your asset and dam management stuff. You've got more software already and we haven't even, we're 28 minutes into an episode.

Brian: [00:29:19] You're talking about the progress machine. Where does the progress machine lead to? That's I think kind of what you're asking is like, what's the point of all this?

Phillip: [00:29:29] Always. It's basically a [00:29:31] rule. There will always be a middle layer of infrastructure that's required and a middle layer of management to manage that infrastructure that's required. [00:30:43] In five years' time we're going to look back and say, "Wow, this didn't actually reduce any amount of capital input. It actually just moved it to a different place in the economy, to a different style of job."

Brian: [00:30:54] But for the people that are out in front of it, there is an arbitrage opportunity. That's exactly what it is.

Phillip: [00:31:01] That's what people that listen to this show are all after. They all want to know how they can use it in their business so they can get some efficiency or arbitrage right now. But I have never seen technology adopted at this rate or become so recognizable. So this is the other thing...

Brian: [00:31:16] The flip side is if you don't adopt it, you get left behind.

Phillip: [00:31:20] You may get left behind, but there's always going to be someone who's going to be like, "I can recognize machine-generated essays and machine-generated output so quickly." I'm already getting hip to it. It took me a couple of months. It was extremely impressive, like a magic trick at the beginning. And then I realized after working with ChatGPT for two months...

Brian: [00:31:41] There's a cadence to this.

Phillip: [00:31:42] You can sort of catch it now.

Brian: [00:31:44] Although if you have a good, as you said, prompt engineer and prompt editor, a human partner, it will be indistinguishable because it will be a human output.

Phillip: [00:31:54] Someone said at some point, "Forgive the length of my letter, if I had had more time, it'd be much shorter." I butchered that quote.

Brian: [00:32:02] Yeah, that's a good quote.

Phillip: [00:32:03] That's how I feel about ChatGPT and other machine-assisted outputs. Now it's like people are so impressed by the length of prose that they just take it all and they shove it all together. They'll ask seven prompts, they'll get 100, I don't know, they'll get 1000 words or 1500 words. They'll just mangle them all together and they'll try to glue the seams together and say, "Look at this thing I just wrote."

Brian: [00:32:26] Yeah.

Phillip: [00:32:27]  [00:32:26]But I don't have the attention span to read that, never mind write it. And that's part of the telltale sign of this output at the moment is the volume of the output is the thing that people are sort of clued into as like, "I can make 1000 word posts," but in reality it requires...  [00:32:43]

Brian: [00:32:43]  [00:32:41]I think you're dead on.

Phillip: [00:32:43]  [00:32:41]It requires a human to output a 400 word post. [00:32:46]

Brian: [00:32:47]  [00:32:47]That's correct. Yeah. And actually condensed writing is one of the most difficult skills of all time. [00:32:53]

Phillip: [00:32:54]  [00:32:54]Yeah. Say that, but shorter. [00:32:54]

Brian: [00:32:54]  [00:32:54]"Say that, but shorter" is the thing that would have killed Dickens but is important especially in business writing. So it's interesting like what will be the markers of when something is written by a human. [00:33:09]

Phillip: [00:33:09]  [00:33:09]Here's a couple of things. A marker right now of what's not generated by ChatGPT is anything that has recency, so meme-ified language, things that are happening right now on Twitter, even using sort of like Gen Z slang, it seems really hard to get the machine to output that sort of thing. So there's like if you know, you know quality and sort of referencing current events, those are things that it just can't do at the moment. And so if you want to have this authenticity layer, maybe dressing up the machine output with a bunch of in-jokes of the moment. [00:33:44]

Brian: [00:33:44] There's a ways around that though, because especially with ChatGPT, you can give it context and feed it information.

Phillip: [00:33:50] It theoretically will tell you that it has no knowledge of events after middle of 2022 or early 2022. So anyway, let's come back to one of the things that I'm trying to tease out here, which is the future of commerce. I don't think necessarily that the channels that we shop in right now are going to be dramatically different in the next year or two. But I want to look forward as to what might be possible in the next five years. We talked about this at our New York salon. So you had posed this question in the Quantum Yeet article about needing new levels of compute and storage to handle the world's information in an explosion of information as we enter into the next age. And it's not just about the capture of new data creation and information. For you, you were talking about context. Also capturing the context in which that information was created is important for later analysis so that we can have this godlike ability to know what the customers journey truly was across all these sales channels.

Brian: [00:34:52] So the data about the data, about the data, about the data.

Phillip: [00:34:55] Right. It's a rabbit hole. So it's very meta. So but here I'm going to I kind of posited to you, well, what if we thought about this differently? It doesn't have to be quantum computing, but what if and maybe this is like...

Brian: [00:35:12] Let's go down this hole. I love it.

Phillip: [00:35:14] So the reason that computing has to follow Moore's Law, which Moore's Law was an exponential, ever-increasing climb in computing power since the 1970s, and sort of that transistor density doubles every 18 months. It's wild that after 35, 40 years, it still holds, but that's kind of where we are. And so we've had to create these engineering capabilities to kind of hold true to this need of ever more computing power. We're now kind of at theoretical limits in that we can only get transistors so close together before we're on a molecular scale. So we're at like three nanometer processing at the moment. We're kind of at the physical limit of a silicon transistor. We can't go any further. So your posit was Quantum is the next realm. But I have another theory. Here's what I think could be possible. We build human interfaces to build computers, so it needs to be human understandable. So right now, you're going to have to follow this journey and it's going to be a little bit of a dissertation, but here we go. The closest thing that runs on a CPU is something called machine code. And machine code, typically to be most performant is written in very almost human illegible code. It's assembler. It's something that's extremely low level. It doesn't need to be interpreted by an operating system. It's literally running on the processor. It's code that's runnable by a transistor and instruction set for a CPU. Okay. On top of that, we have operating systems which handle all kinds of generalized compute functions. It's RAM and memory management, it's storage and hard drive storage. It's Ethernet connectivity and data packets and error correction. There is right now in your CPU a function which is doing error correction for gamma rays that are floating through the atmosphere right now that are coming from outside distant galaxies and supernova that happened billions of years ago. It is fascinating the kinds of things that your computer does that you're not even aware of doing. But a lot of that is written on human interfaces. That's what your operating system is doing at this exact moment. On top of that, you have applications and that's written in C, very performant code, but it's human written for human consumption. On top of that, you have applications which are written in C++. Your web browser is written in C++. It's extremely performant, but it's written in human-legible code. The website that you're accessing is talking to a web server, also written in C++, written in human understandable code. On top of that, we have HTML and CSS and JavaScript and now we have webassembly. And on top of that we have visual display languages and we have JavaScript is actually kind of terse. [00:38:10] So we've created things like React and we have data interchange formats like GraphQL and all of this. All these layers upon layers, upon layers, upon layers is meant for human consumption and has an efficiency cost associated with it. So what if none of that's necessary. And what if you're just talking to a computer and you say, "I need a web page that shows you the Archetypes Journal," that you can buy right now at ArchetypesJournal.com, "and I want that web page to sell things to people, and I want it to go to my bank account." And I can, in exacting detail, give every single instruction of how the visual design and how the interchange works between my bank. What if all of that runs at the lowest level of the CPU? It doesn't have to run on an operating system. It's a purpose built app that runs as close to bare metal as possible. We recapture 40 years, four generations worth of compute power back to humanity. [00:39:14]

Brian: [00:39:14]  [00:39:15]Of workarounds. Basically of workarounds. [00:39:16]

Phillip: [00:39:16]  [00:39:16]It's all workarounds. It's all layers on layers. [00:39:19]

Brian: [00:39:19] Because those are jobs that should be done. There should be one job for humans that interact with machines and the rest of it should be machine to machine interaction. I think you've nailed it.

Phillip: [00:39:31] We don't need quantum. We can buy ourselves another few decades of Moore's Law by just recapturing all of the spare compute power we have laying around.

Brian: [00:39:38] The quantum computing becomes a future generation problem.

Phillip: [00:39:43] Yeah, and we can continue down this path of just human interface connections.

Brian: [00:39:47] Data storage is still the issue, actually.

Phillip: [00:39:50] I think it is. That is a problem.

Brian: [00:39:51] Yeah.

Phillip: [00:39:51] We could cover that in a future episode of Future Commerce. Brian, any last thoughts to kind of wrap us up?

Brian: [00:39:57] My gosh. [00:39:58] Well, let me just read you something that was from 1964 really quickly. This Norbert Wiener, and in a different book actually that I bought of his and he says he was talking about how good can this actually get? He was talking about how language models and he was forecasting like translation of languages from one language to another. And he said if you can get a computer to take a language, translate it into another language and to another language and then back to the original language, you've won. You've won the battle because that's about as complex as it gets. And he said that it doesn't look like in the near term, and again, this is written in 1964, that we're going to be able to get to a point where you can run that circle in full based off of just training an algorithm to do it. And so he said, "It seems to me that the best hope of a reasonably satisfactory mechanical translation is to replace a pure mechanism, at least at first, by a mechanical human system involving as critic and expert human translator, to teach it by exercises as a school teacher instructs human pupils. Perhaps at some later stage, the memory of the machine may have absorbed enough human instruction to dispense with later human participation, except perhaps for refresher course now and then. In this way the machine would develop linguistic maturity." I just think this is dead on for the way that we have to interact with ChatGPT at this moment and as we look ahead, this is interaction with human is the training model for how AI will evolve into what it needs to be to become what you just talked about. It won't be able to do it until it spends enough time interacting with humanity and we're going to have to lead it there. [00:41:57]

Phillip: [00:41:57] We're going to have to leave it right there. There's so much. I have so many more things I want to say. Thank you so much for listening to this episode of Future Commerce. Maybe far Future Commerce at this point. We have more episodes of this podcast and all Future Commerce properties at FutureCommerce.fm, and if you want to stay up to date on when our next salon is going to be, we'd love to have you in attendance and we're working on some really fun things right now. We'd love for you to be involved. Join our community at FutureCommerce.fm/Subscribe. We're in your inbox once, twice, maybe three times a week on a great week. And we'd love to get to know you and have you join our community of futurists. Thank you so much for listening to Future Commerce.

Brian: [00:42:36] Thanks.