My memory of AOL Time Warner is that AOL had the higher valuation at the time and basically bought Time Warner with funny money. It seemed absurd... part of the big tech stock bubble.
I don’t know about the Sears merger, but aren’t you retconning the AOL/Time Warner merger with more respect than it had? I remember it being the target of universal mockery - AOL was already a dinosaur by then, and people couldn’t imagine what anyone would want with it. Sort of like Musk buying Twitter. By the time he bought it it was already a moldering corpse.
The bank branch within the grocery store became a staple of suburban retail for 25 years or so, say 1990-2015, so Sears wasn't completely insane. However, the grocery store sublet worked because it allowed people to combine two very regular errands into one, whereas how often do people find themselves in the market for a new lawnmower and an IRA at the same time?
FWIW, I've been using Duolingo to learn languages for about 10 years now so it will be very interesting to see how their new Duolingo Max upgrade that will use GPT-4 enabled role play and an "explain the answer" feature will fare. https://blog.duolingo.com/duolingo-max/ Seems like perhaps a confirmation and refutation of your thesis. The action is in the specific app but apparently GPT-4 's schmuck (such vulgarity, honestly!) plays a role as well. Duolingo is already publicly traded (NASDAQ: DUOL, market cap about US$10B) so it's not a venture capital prospect, but its future performance might tell an AI story. Education, licensing, certification, and proficiency attestation generally may offer venture capital opportunities, particularly in the nonprofit sector. Hundreds, if not thousands, of nonprofits are in the credentialling racket. The typical model is selling overpriced textbooks and then arranging for proctored testing at the local community college. AI enabled overhauls of such programs might offer ROI by increasing completion rates which would encourage increased enrollment. A huge schmuck is not necessary in such situations, but if the huge schmuck can be more cheaply trained to excel in a particular use case, there may still be advantages to it. However, since so much internet knowledge is irrelevant at best, maybe the ability to come up with a constrained training set of richer, deeper, more accurate, specifically targeted training data to use in such apps will be cost effective if for no other reason than operating energy consumption.
for the collected works of classic authors whose work is now in the public domain. Would they sell even better bundled with a chatbot app trained exclusively on them along perhaps a small, highly curated collection of good quality critical interpretations? Or imagine the public domain Harvard Classics sold with a similar chatbot app offering comprehension and retention tests and tools but completely uncontaminated by internet garbage. It might be a godsend for homeschoolers or autodidacts generally. I would love to see someone like Larry Seidentop sell his ebooks bundled with Guizot, Tocqueville, and Fustel de Coulanges, some lecture videos and a chatbot with comprehension testing. It could be the successor technology to The Great Courses. And it would likely improve the quality of public discourse by a non-negligible amount. Some high minded venture capitalist ought to take a flyer: it could be next big thing in entertainment and leisure.
Fully agree with the points that it won't be "one model" and that the human form factor will not be "the one" - though of course there may be some important use cases where a human form will be ideal.
1 Your medical diagnosis example is far more compelling than the kitchen example you used previously. Doesn't mean all uses will be specialization. I think it is a guess as to which will eventually be more common.
2 Your examples of supermarket are awful. Sears and AOL were failures independent of their combining with DW and TW. Not all banks have expanded services but Chase has brokerage in their branch bank offices. Microsoft was an operating system that became Office software, web browser, phones (failed?), cloud services, etc. Amazon was a book seller (mostly third party) that became a consumer goods seller that now has grocery stores, a delivery system, and probably other major components I'm not thinking of or don't know about. Lots of other big examples. Lots of other smaller examples that we don't really even think about.
Yes, but I'm pretty sure their consumer businesses are mostly profitable. Very different from Sears and AOL. And most or all are in the supermarket theme.
Let's start with Kling. Giving two examples of attempts at superstore that failed does not in any way mean the model does not work. Period.
Regarding my hindsight bias, I didn't verify my memory like I should have but I don't think I was wrong. (Yes, I understand Sears was successful for near a century before it wasn't.) Wikipedia agrees with me that Sears peaked in the 70s. Lots more in there it seems you don't know.
I suppose the argument could be made they peaked long before the 70s but surely not later. My memory is they bought DW and Coldwell Bank, and then Advance auto and others as last gasps. They sold DW with Discover in the late 80s about the time they sold the Sears tower. Walmart surpassed them in about 91 but the slide started long before.
Beyond what I've already said here, your facts are quite a bit off. It is true Discover was a success but as I mentioned, they sold it quite early so I don't even know why you mention it. Sears simply was not doing fine in the 80s. They were already a shrinking company trying to stay relevant by buying other businesses.
Sears' superstore failed because it failed as a consumer goods retailer, the heart of its superstore. AOL TW failed because AOL's primary product became obsolete. Neither had anything to do with whether a superstore is a good idea or not.
To clarify, I'm not saying a superstore is a good model, just that these are bad examples against it.
My memory of AOL Time Warner is that AOL had the higher valuation at the time and basically bought Time Warner with funny money. It seemed absurd... part of the big tech stock bubble.
It was the case of the schmuck wagging the dog.
I don’t know about the Sears merger, but aren’t you retconning the AOL/Time Warner merger with more respect than it had? I remember it being the target of universal mockery - AOL was already a dinosaur by then, and people couldn’t imagine what anyone would want with it. Sort of like Musk buying Twitter. By the time he bought it it was already a moldering corpse.
The bank branch within the grocery store became a staple of suburban retail for 25 years or so, say 1990-2015, so Sears wasn't completely insane. However, the grocery store sublet worked because it allowed people to combine two very regular errands into one, whereas how often do people find themselves in the market for a new lawnmower and an IRA at the same time?
FWIW, I've been using Duolingo to learn languages for about 10 years now so it will be very interesting to see how their new Duolingo Max upgrade that will use GPT-4 enabled role play and an "explain the answer" feature will fare. https://blog.duolingo.com/duolingo-max/ Seems like perhaps a confirmation and refutation of your thesis. The action is in the specific app but apparently GPT-4 's schmuck (such vulgarity, honestly!) plays a role as well. Duolingo is already publicly traded (NASDAQ: DUOL, market cap about US$10B) so it's not a venture capital prospect, but its future performance might tell an AI story. Education, licensing, certification, and proficiency attestation generally may offer venture capital opportunities, particularly in the nonprofit sector. Hundreds, if not thousands, of nonprofits are in the credentialling racket. The typical model is selling overpriced textbooks and then arranging for proctored testing at the local community college. AI enabled overhauls of such programs might offer ROI by increasing completion rates which would encourage increased enrollment. A huge schmuck is not necessary in such situations, but if the huge schmuck can be more cheaply trained to excel in a particular use case, there may still be advantages to it. However, since so much internet knowledge is irrelevant at best, maybe the ability to come up with a constrained training set of richer, deeper, more accurate, specifically targeted training data to use in such apps will be cost effective if for no other reason than operating energy consumption.
There seems to be a good e-book
for the collected works of classic authors whose work is now in the public domain. Would they sell even better bundled with a chatbot app trained exclusively on them along perhaps a small, highly curated collection of good quality critical interpretations? Or imagine the public domain Harvard Classics sold with a similar chatbot app offering comprehension and retention tests and tools but completely uncontaminated by internet garbage. It might be a godsend for homeschoolers or autodidacts generally. I would love to see someone like Larry Seidentop sell his ebooks bundled with Guizot, Tocqueville, and Fustel de Coulanges, some lecture videos and a chatbot with comprehension testing. It could be the successor technology to The Great Courses. And it would likely improve the quality of public discourse by a non-negligible amount. Some high minded venture capitalist ought to take a flyer: it could be next big thing in entertainment and leisure.
Fully agree with the points that it won't be "one model" and that the human form factor will not be "the one" - though of course there may be some important use cases where a human form will be ideal.
1 Your medical diagnosis example is far more compelling than the kitchen example you used previously. Doesn't mean all uses will be specialization. I think it is a guess as to which will eventually be more common.
2 Your examples of supermarket are awful. Sears and AOL were failures independent of their combining with DW and TW. Not all banks have expanded services but Chase has brokerage in their branch bank offices. Microsoft was an operating system that became Office software, web browser, phones (failed?), cloud services, etc. Amazon was a book seller (mostly third party) that became a consumer goods seller that now has grocery stores, a delivery system, and probably other major components I'm not thinking of or don't know about. Lots of other big examples. Lots of other smaller examples that we don't really even think about.
Yes, but I'm pretty sure their consumer businesses are mostly profitable. Very different from Sears and AOL. And most or all are in the supermarket theme.
Let's start with Kling. Giving two examples of attempts at superstore that failed does not in any way mean the model does not work. Period.
Regarding my hindsight bias, I didn't verify my memory like I should have but I don't think I was wrong. (Yes, I understand Sears was successful for near a century before it wasn't.) Wikipedia agrees with me that Sears peaked in the 70s. Lots more in there it seems you don't know.
https://en.m.wikipedia.org/wiki/Sears#:~:text=Sears%20reached%20its%20pinnacle%20in,the%20Sears%20Tower%20in%201988.
I suppose the argument could be made they peaked long before the 70s but surely not later. My memory is they bought DW and Coldwell Bank, and then Advance auto and others as last gasps. They sold DW with Discover in the late 80s about the time they sold the Sears tower. Walmart surpassed them in about 91 but the slide started long before.
Beyond what I've already said here, your facts are quite a bit off. It is true Discover was a success but as I mentioned, they sold it quite early so I don't even know why you mention it. Sears simply was not doing fine in the 80s. They were already a shrinking company trying to stay relevant by buying other businesses.
Sears' superstore failed because it failed as a consumer goods retailer, the heart of its superstore. AOL TW failed because AOL's primary product became obsolete. Neither had anything to do with whether a superstore is a good idea or not.
To clarify, I'm not saying a superstore is a good model, just that these are bad examples against it.
https://twitter.com/coreylynch/status/1767927194163331345
I’m updating my prior to favor general purpose humanoid robots able to do just about anything a human could do, tho maybe a bit better.
The multi-armed super bartender is coming! Along with more specialized & limited automatic service. And even less human contact, more loneliness.
To let each user subspecialize the training.