For a product to be able to serve billions of people at an ultra low cost (free), revenue must come from somewhere, and ads are a great way. Taste and boundaries will matter and based on many lessons from Instagram and tiktok, ads have a higher chance of being introduced in an non invasive, surprisingly value added way.
I pay $140 per year for YouTube Premium in order to avoid ads. Of the 2.5 billion active users about 5% pay for this service. The service does come with some add ons like YouTube Music. What I object to is the embedded adds in videos for "sponsors". One of the most egregious is Square Space, who seem to sponsor the world. With the advent of AI website development tools I suspect they will drift to the background. The presence of embedded ads usually determines whether I decide to subscribe to a channel. It sounds as though the ads on AI services will sit along side content and not be as intrusive. We've lived with ads on the internet almost from the start and should not expect any difference for an otherwise free service.
Thanks for this thoughtful piece, Bryan. I certainly agree with the headline "Of course they’re putting ads in AI". As an early consumer Internet pioneer, I get it- advertising is a powerful monetization model, but I was hoping for something better this time around. In many ways, what we did 30 years ago was to take the print and TV media business model and adapt it with some interactivity to the Internet. This AI moment represents an opportunity to rethink how humans interact with media and technology. My gut tells me we should spend more time and be more creative about how we implement consumer business models instead of just defaulting to advertising.
Yes would love that- it’a still in the “ads” category to some extent but one of the ideas I added was a personally set goal based bidding where individuals set a price / bounty on tbe results they want to see from AI.
I’m OK with ads as long as it’s in the free tier. What I’m not ok with is paying the subscription and still getting ads like all the major streaming platforms.
We got the frontier models with incredible intelligence. They unfortunately come with some marginal cost due to inference / compute, so in order to serve billions...the bill can be footed by many (investors, consumers and...other companies via ads!)
Super open to it! I just think of the source of revenue.. users can pay, the companies can pay (via subsidy from investor dollars or sell user data) or someone else needs to foot the bill.
This is a literal physics and math problem. It’s about energy, the use, transference, latency, distribution… just like how CPU / GPU / NPU’s are now stacked in modern (macOS) architecture and how we kind of walked our way back into the solution for LLMs via video gaming. Surprise.
"internet has always run on ads and most people like it that way" ... big statement, but not completely true. People DO NOT like it that way, but people have no choice. The same happens with the Privacy choices to access websites; you have to give up your data in order to access content.
Ads would happen in A no matter how many people protest them. However, the toxic way they will be introduced when you are searching for an article or doing "deep research" is what will make it unbearable.
Thank you for the opening to have an intelligent discussion about the merits of the current monetization model put forth by a few companies. As someone who has been playing around with computers since the early 80s and can still remember the joy and sense of power I felt when I was able to upgrade from a 9600 Baud modem to a 14.4K modem. I can say emphatically that the internet has not always run on ads. Prior to Hotmail and Gmail, we paid for our email addresses. We paid for the CompuServe or AOL CD that would let us get on to the internet. Even the current monetization model banks on the fact that people do not like ADs. That's why they charge more money for Ad-Free options. If people liked Ads, that model would not work. You observation that heavier users of the LLMs gladly pay for higher tier subscriptions further supports this. The current monetization model is essentially, "we will introduce something you don't like to influence you to pay for the service." I currently pay for subscriptions to the big 3 providers and as such wouldn't be effected by Ads in free tiers. My concern is how much longer until the recommendations or information we receive from these models are influenced by the amount of money that has been offered? This is something I think is important to get a handle on before the majority of the conversations are happening via autonomous agents and we as the end user, won't have as much visibility into any biases.
Customers vote with their wallets for products and services they love and value most. For everything else, they (users, not customers) settle for free-with-ads until something better comes along.
For a product to be able to serve billions of people at an ultra low cost (free), revenue must come from somewhere, and ads are a great way. Taste and boundaries will matter and based on many lessons from Instagram and tiktok, ads have a higher chance of being introduced in an non invasive, surprisingly value added way.
I pay $140 per year for YouTube Premium in order to avoid ads. Of the 2.5 billion active users about 5% pay for this service. The service does come with some add ons like YouTube Music. What I object to is the embedded adds in videos for "sponsors". One of the most egregious is Square Space, who seem to sponsor the world. With the advent of AI website development tools I suspect they will drift to the background. The presence of embedded ads usually determines whether I decide to subscribe to a channel. It sounds as though the ads on AI services will sit along side content and not be as intrusive. We've lived with ads on the internet almost from the start and should not expect any difference for an otherwise free service.
Again, someone needs to foot the bill! AI is not a zero marginal cost product.
Thanks for this thoughtful piece, Bryan. I certainly agree with the headline "Of course they’re putting ads in AI". As an early consumer Internet pioneer, I get it- advertising is a powerful monetization model, but I was hoping for something better this time around. In many ways, what we did 30 years ago was to take the print and TV media business model and adapt it with some interactivity to the Internet. This AI moment represents an opportunity to rethink how humans interact with media and technology. My gut tells me we should spend more time and be more creative about how we implement consumer business models instead of just defaulting to advertising.
Yes would love that- it’a still in the “ads” category to some extent but one of the ideas I added was a personally set goal based bidding where individuals set a price / bounty on tbe results they want to see from AI.
I’m OK with ads as long as it’s in the free tier. What I’m not ok with is paying the subscription and still getting ads like all the major streaming platforms.
Yes hard agree. Paying for sub then seeing an ad feels like a breach of contract
“We were promised flying cars… and we got 140 chars…” — Peter Thiel
I really believe AI and the application of it can produce real, distinct, novel implementations that we have not considered before.
So, ads are fundamentally disappointing on so many levels.
We got the frontier models with incredible intelligence. They unfortunately come with some marginal cost due to inference / compute, so in order to serve billions...the bill can be footed by many (investors, consumers and...other companies via ads!)
Your view is still too acute.
You’re presuming the present biz models are still the best (to support this novel tech).
A new biz model clearly must emerge. Of course, this was the promise of bitcoin and crypto too.
Super open to it! I just think of the source of revenue.. users can pay, the companies can pay (via subsidy from investor dollars or sell user data) or someone else needs to foot the bill.
This is a literal physics and math problem. It’s about energy, the use, transference, latency, distribution… just like how CPU / GPU / NPU’s are now stacked in modern (macOS) architecture and how we kind of walked our way back into the solution for LLMs via video gaming. Surprise.
Then come the economics.
"internet has always run on ads and most people like it that way" ... big statement, but not completely true. People DO NOT like it that way, but people have no choice. The same happens with the Privacy choices to access websites; you have to give up your data in order to access content.
Ads would happen in A no matter how many people protest them. However, the toxic way they will be introduced when you are searching for an article or doing "deep research" is what will make it unbearable.
Thank you for the opening to have an intelligent discussion about the merits of the current monetization model put forth by a few companies. As someone who has been playing around with computers since the early 80s and can still remember the joy and sense of power I felt when I was able to upgrade from a 9600 Baud modem to a 14.4K modem. I can say emphatically that the internet has not always run on ads. Prior to Hotmail and Gmail, we paid for our email addresses. We paid for the CompuServe or AOL CD that would let us get on to the internet. Even the current monetization model banks on the fact that people do not like ADs. That's why they charge more money for Ad-Free options. If people liked Ads, that model would not work. You observation that heavier users of the LLMs gladly pay for higher tier subscriptions further supports this. The current monetization model is essentially, "we will introduce something you don't like to influence you to pay for the service." I currently pay for subscriptions to the big 3 providers and as such wouldn't be effected by Ads in free tiers. My concern is how much longer until the recommendations or information we receive from these models are influenced by the amount of money that has been offered? This is something I think is important to get a handle on before the majority of the conversations are happening via autonomous agents and we as the end user, won't have as much visibility into any biases.
Customers vote with their wallets for products and services they love and value most. For everything else, they (users, not customers) settle for free-with-ads until something better comes along.