Which do you think is more important? Putting man on the moon or ecommerce? I reckon you been able to get on a device, see a biscuit ads, order one from foo.com and have it shipped to you. Think of how much tech it takes for that to happen, that is more tech than NASA built to send many to the moon, the internet, packet switching, routing, fiber optic, distributed systems, web servers, web browsers, ads, cryptography, banking online, and so on and so forth. We love to trivialize what is common, but that clicking on an ad is not an easy problem. Clicking on ads has generated enormous wealth in the world which is now bootstrapping AGI.
Clicking on ads helped with our goal to AI today. Showing you the right ad and beating those trying to game it is machine learning heavy. When was the first time we started seeing spelling correction and next word suggestions? It was in google search bar. To serve the correct ads and deal with spam? heavy NLP algorithms. If you stop and think of it, we can drop a think line from the current state of LLMs to these ads click you are talking about.
It took way too long to convince myself this wasn't satire. I still wish it wasn't.
It made me realize that I think many computing people need more of a fundamental education in "hard" physics (statics, mechanics, thermodynamics, materials science) in order to better understand the staggering paradigm shift that occurred in our understanding of the world in the early 20th century. Maybe then they would appreciate how much of the world's resources have now been directed by the major capital players towards sucking the collective attention span of humanity into a small rectangular screen, and the potential impact of doing so.
The comparison here is between moonlanding and advertisement. So I choose the moon obviously.
Ecommerce can work just the same without LLM augmented personalized ads, or no advertisement at all. If a law would ban all commercial advertisement - people still need to buy things. But who would miss the ads?
To complement your argument, I once was invited to speak at an event in Sao Paolo, Brazil, and was astonished to learn that outdoors advertising is _banned_ in that city.
Yet somehow people have still been buying mattresses, cars, and laundry detergent.
I can't say I missed the ads one minute during my trip.
They are clearly talking about one aspect of the industry which is the marketing part related to maximising engagement. It is not meant to be conflated with the e-commerce industry as a whole.
Interesting. In my experience, advertisement and the incentives around it have led to the most devastatingly widespread removal of value in human culture and social connections that we've seen in this generation. Huge amounts of effort wasted on harvesting attention, manipulating money away from people, isolating and fostering extremism, building a massive political divide. And centralizing wealth more and more.
The amount of human effort wasted on advertisement is staggering and shocking.
I don't think your average adult is inspired by the idea of AI generated advertisements. Probably a small bubble of people including timeshare salesmen. If advertisements were opt-in, I expect a single digit percentage of people would ever elect to see them. I don't understand how anybody can consider something like that a net good for the world.
How does non-consensually harassing people into spending money on things that don't need add value to all the world's citizens?
"Adding value" and "Generating wealth" are always the vague euphemisms that these guys fall back to when they try to justify much of today's economic activity. Adding value for who? Generating whose wealth? The answer is usually "people who are already wealthy." Of course, they'll downplay the massive funneling of wealth to these people, and instead point to the X number of people "lifted out of poverty in the 20th century" as if capitalism and commerce was the sole lifting force.
I wish some of these people would think about how they'd explain to their 5 year old in an inspiring way what they do for a living: And not just "I take JSON data from one layer in the API and convert it to protobufs in another layer of the API" but the economic output of their jobs: "Millions of wealthy companies give us money because we can divert 1 billion people's attention from their families and loved ones for about 500 milliseconds, 500 times a day. We take that money and give some of it to other wealthy companies and pocket the rest."
> If advertisements were opt-in, I expect a single digit percentage of people would ever elect to see them.
I mean, you'd see the same thing if paying for your groceries were opt-in. Is that also a net bad for the world? Ads do enable the costless (or cost-reduced) provision of services that people would otherwise have to pay for.
Ads are not charity. There is clearly a cost, otherwise they would lose money. They do not generate money out of thin air. "Generate" and "extract" aren't synonyms.
They do not enable any costless anything at all. They obfuscate extraction of money to make it look costless, but actually end up extracting significant amounts of money from people. Ad folks whitewash it to make it sound good, but extracting money in roundabout ways is not creating value.
> you'd see the same thing if paying for your groceries were opt-in.
Groceries are opt-in. Until you realize you don't want to hunt and cook your own food, then you opt back in for survival.
UBlock origin + some subscriptions show I'd definitely would love to opt out of IRL ads.
>Is that also a net bad for the world?
World, yes. We have to tech to end food scarcity, but poor countries struggle while rich countries throw out enough food each day to feed said poor countries.
I think this is a rationalization of an enormous waste of work. The effects generating wealth are indirect. In that regard you could argue that betting is generating wealth too. Advertising is like a hamster wheel people have to jump onto if they want their place in the market.
A similar amount of wealth would be generated if every advertised product would be represented by a text description, but we have a race to the bottom.
There is advertising and advertising of course but most of advertising is incredibly toxic and I would argue that by capturing attention, it is a huge economic drain as well.
Of course an AI would also be quite apt at removing unwanted ads, which I believe will become a reality quite soon.
> A similar amount of wealth would be generated if every advertised product would be represented by a text description, but we have a race to the bottom.
I fear statements like this go too far. I can't agree with the first part of this sentence.
I feel this about both marketing and finance:
They are valuable fields. There are huge amounts of activity in these fields that offer value to everyone. Removing friction on commerce and the activities that parties take in self-interest to produce a market or financial system are essential to the verdant world we live in.
And yet, they're arms races that can go seemingly-infinitely far. Beyond any generation of societal value. Beyond needless consumption of intellect and resources. All the way to actual negative impacts that clutter the financial world or the ability to communicate effectively in the market.
In the grand scheme, what you’re talking about is very zero-sum, while stuff like making rockets is not. Uber vs Waymo is a good example of how adtech can only go so far in actually creating wealth.
I keep hearing the phrase "generate wealth" in regards to advertisement and from the mouths of startup founders, but in almost no other context. I'm not familiar with the economic concept of "wealth generation" or its cousin "creating value".
Is the idea that any and all movement of money is virtuous? That all economic activity is good, and therefore anything that leads to more economic activity is also good? Or is it what it sounds like, and it just means "making some specific people very wealthy"? Wouldn't the more accurate wording be that it "concentrates wealth"? I don't see a huge difference in the economic output of advertisement from most other scams. A ponzi scheme also uses psychological tricks to move money from a large amount of people to a small amount of people. Something getting people to spend money isn't inherently a good thing.
> Is the idea that any and all movement of money is virtuous?
Maybe this was your point, but this is built in to one of the definitions of GDP, isn’t it? Money supply times velocity of money?
I’m no economist though I’m sure there are folks on here who are. But this seems like an unfortunate fact that’s built into our system- that as laypeople we tend to assume that ‘economic growth’ means an increase in the material aspects of our life. Which in itself is a debatable goal, but our GDP perspective means even this is questionable.
For example, take a family of five living out in a relatively rural area. In scenario one, both parents work good paying remote tech jobs and meals, childcare, maintenance of land and housing, etc. are all outsourced. This scenario contrubutes a lot according to our economic definitions of GDP. And provides many opportunities for government to tax and companies to earn a share of these money flows.
Then take scenario 2, you take the same family but they’re living off of the grid as much as possible, raising or growing nearly all their own food, parents are providing whatever education there is, etc. In this scenario, the measurable economic activity is close to zero- even if the material situation could be quite similar. Not to mention quality of life might be rated far higher by many.
What rating an economy by the flow of its money does do is, and I’m not sure if this is at all intentional, is it does paint a picture of what money flows are potentially capturable either by government taxation or by companies trying to grab some percentage as revenue. It’s a lot harder to get a share of money that isn’t there and/or not moving around.
Perhaps my take on economics is off base but, for me, seeing this made me realize just how far off our system is from what it could and should be.
GDP is a measure. I'm very much not an economist, but I am extremely skeptical that the health of an economy can be reduced to any single number. Goodheart's law and all.
I concede that GDP is a good indicator, but I think you can have things that help GDP while simultaneously hurting the economy. Otherwise any scam or con would be considered beneficial, and it would make sense to mandate minimum individual spending to ensure economic activity. A low GDP inherently shows poor economic health, but a high GDP does not guarantee good health.
In my mind (noting, again, that I'm no economist), economic health is defined by the effectiveness of allocating resources to things that are beneficial to the members of that economy. Any amount of GDP can be "waste", resources flowing to places where they do not benefit the public. As Robert Kennedy famously pointed out, GDP includes money spent on addictive and harmful drugs, polluting industries, and many other ventures that are actively harmful.[0]
Going back to the previous posters monetary velocity statement, if you have a trillion dollar GDP, but it's just two AI's bouncing money back and forth high speed while all the humans starve in the street your economy is "great" and totally awful at the same time. The one number has to be referenced against others like wealth inequality.
"Generate wealth" means "make somebody's number go up" i.e. allocating real resources/capital somewhere, with the assumption that 1. allocating that capital creates a net boon for society and 2. those who have "generated wealth" are wise and competent investors/leaders and their investments will create a net boon elsewhere. The first point is probably not especially true very often in contemporary tech (other than 'job creation') and is arguably not true for advertisement. The second point is not really a given at all and seems to be pretty consistently shown otherwise.
Clicking on ads helped with our goal to AI today. Showing you the right ad and beating those trying to game it is machine learning heavy. When was the first time we started seeing spelling correction and next word suggestions? It was in google search bar. To serve the correct ads and deal with spam? heavy NLP algorithms. If you stop and think of it, we can drop a think line from the current state of LLMs to these ads click you are talking about.