How to Make ChatGPT Ads Not Suck

Today on the AI Daily Brief we are talking about OpenAI's announcement that ads are, as they were always inevitably going to, coming to ChatGPT. The question is, can they make them not suck? And I think I have a few ideas for how they could achieve that. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Alright friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, ZenCoder and Super Intelligent. To get an ad free version of the show go to patreon.com aidaily Brief or you can subscribe on Apple Podcasts. To learn about sponsoring the show or really anything else related to the show, go to aidaily Brief AI. It's there you can find links to sponsorship, to information about speaking engagements or the wide variety of projects that we have going on at any given time, such as our AI New Year's resolution or the upcoming AIDV Intel. Now let's talk about advertising in ChatGPT. On Friday afternoon the official OpenAI account tweeted, in the coming weeks we plan to start testing ads in ChatGPT free and go tiers. We're sharing our principles early on how we'll approach ads, guided by putting user trust and transparency first as we work to make AI accessible to everyone. What matters most responses in ChatGPT will not be influenced by ads. Ads are always separate and clearly labeled. Your conversations are private from advertisers. Plus, pro, business and enterprise tiers will not have ads. They also show an example of someone asking about simple but authentic Mexican ideas for their dinner party, which led to a hot sauce, which led to a Harvest Groceries ad focused on hot sauce. Now in their announcement post they really focus on the fact that advertising is their way to make sure that access to hyper intelligent AI assistance remains available to everyone. They write AI is reaching a point where everyone can have a personal super assistant that helps them learn and do almost anything. Who gets access to that level of intelligence will shape whether AI expands opportunity or reinforces the same divides. So then they say in the long run, ads are what's going to make OpenAI able to continue to provide lots and lots of otherwise free users with high quality service. They expand a little bit what they say about their ads principles from the tweet. Some of them are repeated like answer independence and conversational privacy. But they also articulate a principle of mission alignment. Our mission, they write, is to ensure AGI benefits all of humanity. Our pursuit of advertising is always in support of that mission in making AI more accessible. And alongside that, another principle they articulate is long term value. We do not optimize for time spent in ChatGPT. We prioritize user trust and user experience over revenue. They also hint that despite their examples being extremely simple display ads taking advantage of the high intent of ChatGPT users, maybe they'll think about more creative strategies in the future. They write. Given what AI can do, we're excited to develop new experiences over time that people find more helpful and relevant than any other ads. Conversational interfaces create possibilities for people to go beyond static messages and links. For example, they say soon you might see an ad and be able to directly ask the questions you need to make a purchase decision. They also note ads can also be transformative for small businesses and emerging brands trying to compete. AI tools level the playing field even further, allowing anyone to create high quality experiences that help people discover options they might have never found otherwise. Otherwise. So basically overall they're saying user trust is paramount. We're not trying to turn into the next meta that absorbs all your time. Hopefully we can get creative with new types of ad units, but ultimately this is necessary to provide everyone equal access to high quality artificial intelligence. In his reshare of the OpenAI post, Sam Altman chose to focus on reinforcing the message that ChatGPT ads will not influence the answers that ChatGPT gives you, as well as data privacy but I. E. That advertisers don't get conversational data. CEO of Applications Fiji Simo also focused on that. Most importantly, she writes in her tweet, ads will not influence the answers ChatGPT gives you. So let's talk about responses. For some, this was an inevitable and perhaps necessary evil. Anonymous poster signal actually responded to Sam's posts calling it a necessary evil and then went into a little bit more detail. In their own post they write let's talk about ads, because ads inside ChatGPT will be insane. Meta made about $58 per user in 2025 purely from ads. Now imagine OpenAI hits a billion free users if they monetize at just 9% of Meta's ARPU, that's $5 billion in incremental revenue. 18% gets you $10 billion. Full parity is $57 billion a year from ads alone. Can AI companies ever match or exceed social media? ARPU likely. AI sits closer to intent than feeds ever did. Ads in a feed monetize attention. Ads in an AI convo monetize decisions. It'll take time, but if AI becomes the default interface for thinking, searching, buying and choosing social media. ARPU may end up looking like a floor, not a ceiling. Given that they have many, many Facebook veterans, it looks very likely that OpenAI has the potential to build one of the greatest ad businesses of all time. OpenAI is essentially building Facebook 2.0 and all the old Facebook peeps are doing it, Tanmay writes. My thoughts on OpenAI ads I think it will be a game changer in online advertising. ChatGPT is affiliate marketing on steroids. It grows to know you personally and given context, will tell you exactly what to buy. And if you buy, OpenAI gets a cut. Shared memory across chats will only bolster this. The level of personalization will be unmatched and unlike anything seen in online advertising before. When this rolls out to all users, OpenAI revenue will go ballistic. Now one note showing that these folks might have the right of it. In early studies, it seems very clear and consistent that ChatGPT users have much higher intent compared to Google searchers. One study, for example, found that ChatGPT traffic converted at 16% as opposed to Google Organics 1.76%, a 9x difference in the conversion rate. I've seen many, many other studies, and while that one is on the high end, they all show ChatGPT being higher intent by a meaningful factor than traditional Google search. This makes sense intuitively. If you're asking questions about something in ChatGPT, you're probably doing a more conscientious form of research than just idly googling something. The problem, of course, is that the communication around this has not been good, Meta's Jason Yim wrote in May 2024, Sam Altman said ADS Plus AI is uniquely unsettling to me and called advertising a last resort. Today chatgpt is getting ads. OpenAI is burning through cash at an insane rate. So now the company that positioned itself as the ethical AI alternative to big tech is building an ad platform to monetize its 700 million weekly users. And Sam Altman's new take I love Instagram ads. You can't make this up, Nate Hake writes. So we are all in agreement now that the whole AGI and AI abundance narratives were total scams, right? Benjamin Decracker wrote. Remember just last month OpenAI implied that people were lying about seeing test ads in ChatGPT and that ads were just silly rumors. They were 100% working on ads at the time. Now, interestingly, Ben Thompson from strategery argues that OpenAI should have already put ads in GPT. In a recent interview on TBPN, he said they could have launched the world's crappiest ads in 2023. By today, in 2026, the ads would be good, they'd be making money and people wouldn't rebel against it. Now they're going to have to launch ads and they're going to suck and people are going to be like this sucks. I'll just go to Gemini now. This reflects something that I said back when they launched the Sora app. To anyone watching the numbers, it has been completely inevitable that at some point ChatGPT was going to have to be an ad supported model. It's only something like 5% of users that are converting to paid and you just can't have 95% of users who are using ChatGPT as much as they are and not have an ad based model. It was just completely inevitable. And anyone who has been around any business for any amount of time has known that it is inevitable. Which made it so absolutely patronizing when they pretended that they weren't going to do ads. Remember all of the hubbub around the Sora app was people being frustrated that OpenAI was just turning into another attention hog. The communication around advertising at the time was still oh no, decisions have been made, yada yada yada. So either one the decision had been made and they were lying, or at least not communicating truthfully, or b they were in some sort of state of naive denial that in the last couple of months they finally got themselves out of Neither case is particularly reassuring and you can already see the challenge to trust. Certainly the thing that people are most concerned about is the way that it impacts the quality of results. Sam McRoberts captures the sentiment of about a million different tweets when he says, how can we trust this Ads will not influence the answers ChatGPT gives you. Ads result in an inevitable conflict of interest. Just look at how Google boiled the frog over time. Jason Yim's post goes farther in describing this phenomenon, referring to the new mockups that chatgpt shared. He says the current mockups show Google shopping style ads sitting at the bottom of the screen, clearly separated and not integrated into the actual response. OpenAI is being careful, testing the waters, but here's what history tells us. Google search ads started the same way. Now they're nearly indistinguishable. Just a tiny sponsored label. The playbook is always the same. Introduce ads as separate and non intrusive, wait for users to get used to them, slowly integrate ads deeper into the core experience. Give it 18 months and ChatGPT will be recommending products mid conversation. Based on what you've told me, you might like this with a tiny sponsored tag you'll barely notice. Now I have a different thought on that, as you'll see in a minute. But that is the concern that people have. Other concerns that people have is that even as we're trying to get better memory, this increases the cost of our services having memory. Jen Xu writes, OpenAI launching ads within ChatGPT is the exact reason why I don't want my AI tools knowing about me or remembering my preferences. Yes, ads are annoying, but integrated in ChatGPT, the depth of manipulation could mess you up without noticing. Augustine lebron thinks that this could threaten recruiting. Leave zero sum quant finance and come build the machine. God, they're going to have a rude awakening when it turns out they have to work on ads, ads and more ads to pay for it all. To the extent that there are positives, it's around either a the idea that advertiser pressure could keep ChatGPT within normative boundaries, Business Insider correspondent Katie Natopoulos writes. Here's the upside to ads on ChatGPT. Yes, ads are annoying, but being subjected to advertiser pressure has a normalizing effect. A tech company has to maintain a bare minimum morality or advertisers flee. But still. Overall, most of the responses are somewhere between cynical, skeptical, or outright mad. Alright, let's talk about the signal versus the noise in enterprise AI. The challenge right now isn't just about what's possible, it's about what's practical. That's the entire focus of the you can With AI podcast I host for kpmg. Season one cut through the hype to focus on deployment and responsible scaling. Season two goes a level deeper. We're bringing together panels of AI builders, clients and KPMG leaders to debate the strategic questions that will define what's next for AI in the enterprise. Six episodes packed with frameworks you can actually use. Find you can with AI wherever you get your podcasts. Subscribe now so you don't miss the new season. If you're using AI to code, ask yourself, are you building software or are you just playing prompt roulette? We know that unstructured prompting works at first, but eventually it leads to AI slop and technical debt. Enter zenflow. Zenflow takes you from vibe coding to AI first engineering. It's the first AI orchestration layer that brings discipline to the chaos. It transforms freeform prompting into spec driven workflows and multi agent verification where agents actually cross check each other to prevent drift. You can even command a fleet of parallel agents to implement features and fix bugs simultaneously. We've seen teams accelerate delivery 2x to 10x, stop gambling with prompts, start orchestrating your AI, turn raw speed into reliable production grade output at Zenflow Free. Today's episode is brought to you by Superintelligent. Superintelligent is a platform that, very simply put, is all about helping your company figure out how to use AI better. We deploy voice agents to interview people across your company, combine that with proprietary intelligence about what's working for other companies, and give you a set of recommendations around use cases, change management initiatives that add up to an AI roadmap that can help you get value out of AI for your company. But now we want to empower the folks inside your team who are responsible for that transformation with an even more direct platform. Our forthcoming AI Strategy Compass tool is ready to start to be tested. This is a power tool for anyone who is responsible for AI adoption or AI transformation inside their companies. It's going to allow you to do a lot of the things that we do at superintelligent, but in a much more automated, self managed way and with a totally different cost structure. If you are interested in checking it out, go to aidailybrief AI Compass, fill out the form and we will be in touch soon. So is there a way to make this better or is this just, as Signal put it, a necessary evil? I think ultimately that even without a lot of creativity in the formats, part of the cost benefit analysis of free services on the Internet is the truth in the old maxim that if you are not paying, you're the product. Advertising is simply put the way that services are offered for free. But obviously OpenAI should aspire to more. If for no other reason, then Google is one of the most powerful advertising juggernauts in the history of the world. And so OpenAI really needs to get creative to have some sort of differentiation. So let's turn now away from what has been announced to how I would try to make ChatGPT ads not suck if I were in charge. With the help of Genspark I put together this strategic presentation. OpenAI team, if you are listening, feel free to use this without attribution. Let's just work to make these inevitable ads strive to not only be not painful, but to actually be interesting and good. Now the opportunity and the challenge is obviously that the other juggernaut in the space when it comes to consumer AI has a 20 year head start in Internet advertising. But of course the intense signal that exists in ChatGPT is its own opportunity. And let's be real, there is, simply put, not an option to not do advertising. Even with incredible growth in the annualized revenue all the way up to 20 billion, as per the latest announcements from OpenAI, the company is still burning a huge amount of cash and paid users seem at this point unlikely to be able to cover the gap. We've already discussed the intent advantage, but this really is what makes the potential for advertising in ChatGPT so interesting. Now, I don't want to be cynical about OpenAI's stated principles. I think that those principles are foundational and need to be there. But I think we have to go farther than just stating the principles. I think users need an actual control foundation that gives them granular, legible control that actually empowers them relative to other platforms. Basically, users need to be able to see what the system believes about their preferences and correct it. They need to be able to interact with ads as partners, not targets. When the ads say you seem interested in Japan and serve up some travel tips, the user needs to be able to say I already went. In other words, they need to be able to correct the system. They need to be able to have control of timing and deferrals. I'm interested in this category, but not right now. I'm researching but not buying. Check back in three weeks. The point is to provide easy ways to give users the ability to provide signal that makes the next ad better. Users, of course, don't have to take advantage of that control, but it is there for them if they wish to. But let's take it farther. In addition to just giving them more control, we need a bad ad insurance policy. We need a flag and skip mechanism, and if you see something that is insensitive or mistimed, you need to flag it and get a day of ad free. Basically, ChatGPT needs to put its money where its mouth is when it comes to better ads. I also think in addition to getting ad free days, the company might want to think about transparent advertiser ratings. Give users the ability to rate ads directly, create a consequence with the cost of quality, make it so that low ratings pay more per impression and that higher ratings get preferential access and lower costs and make this all public. Create an advertiser quality rating, show a block rate, show a user satisfaction rate. Again, many people aren't going to take advantage of this, but my guess is that there's a positive correlation between the people who are most concerned about ads disrupting the experience and those who would so with that is our foundations. Let's talk about innovating on the ad units themselves. I think that OpenAI is so concerned right now with not being disruptive that they could very easily miss out on the opportunity to be much more innovative and and potentially high value with the ad units themselves. So here are five different categories of ideas for slightly different approaches to the actual ads. Category one is of course transactional advertising and this is where OpenAI is even starting right now. However, there might be some ways to better align the incentives. Right now, advertising is sold on the basis of potential cost per view or cost per click. That is the pay for attention model. The obvious thing to experiment is to shift that model to pay for results instead to have verified outcomes where the advertiser pays only when the transaction completes and users confirm satisfaction. Think 50 for a booking instead of 2 for a click. Now for those of you who are in advertising out there rolling your eyes because this would seriously diminish the number of advertisers who are able to successfully get value from this because a lot of the products are crappy. That's kind of the point. This would prioritize advertisers who had something valuable to offer where there actually was going to be a verified outcome as opposed to just people who are buying inventory at scale. What's more, we can take advantage of the fact that users of ChatGPT are high intent to actually allow them to shift into a different mode which makes them more open to everything around commerce. Think a user initiated commercial mode. A help me buy button. The old paradigm is interruption. The new paradigm is this as a feature. Think buying agents that actively work for the user negotiating and filtering options with a clear context switch. Here's an example of what some of that transactional advertising might look like. Here a user has asked help me book a flight to Tokyo with a $1,200 budget. ChatGPT responds, Here are two ways I can help either help me book which shows options with a book now button, or an ability to just see options without the advertising integrated. You can also see here that the airlines in question only pay if you complete the booking. Here's another example of being able to switch between just comparing options or actually having assistant signing up with those disclosures around when and how the advertiser pays. Next up, let's talk about offers. Offers are going to be when an advertiser doesn't just serve you up a display ad, but Provide some sort of discount or incentive to go take an action now. Well, maybe we create an offers exchange where in addition to getting served contextual offers in situ as part of conversations, there's also a place where you can go browse the current offers that are available. And for those of you who think there are crazy, there are entire categories of websites that make hundreds of millions, if not billions of dollars a year bas doing things like this, giving people the ability to browse current offers to optimize the timing of their shopping. So I think that this is two parts. The first is offers that are actually useful, taking advantage of the context and intent that they have. Think triggered memory. You said you'd wait for a sale on running shoes, but Nike is 30% off today. Think negotiation. I can get you 25% off if you buy through this partner right now. There could be verified scarcity 47 remaining at this price, basically giving ChatGPT access to real inventory data. There's also price context. This TV is 20% off, but was 25% off last month. Here's the history now, although some of the laments in the wake of the announcement was that people didn't want to see ads in line, if you provide more contextual information that has some of these principles, I think that people could get down with this. Someone asks, I want to get back into running what should I know about building up distance safely. It gives the set of tips and then in a very distinguished ad unit, it says, by the way, six weeks ago you mentioned waiting for a sale on Nike Pegasus 41s. Nike is running 30% off through Sunday. This shows that it's the best price in six months and gives the user a set of contextual options including view the deal, reminding them later or indicating that they are not interested. It also has the ability to immediately adjust your preferences around how you get served these types of offers. But in addition to just showing off in line, the offers exchange could be a place that you actually go to track this. Now, again, some of you are cringing right now because you're thinking Facebook Marketplace. But Facebook Marketplace has become an extremely useful feature for a huge number of users there. And I don't think it's impossible that something similar could happen inside ChatGPT. Number three, let's move on to brand advertising, where instead of funding offers or just transactional ads, brands find capabilities. These are not just performance dollars, these are brand dollars that are going into this. And in this case, brands could more explicitly fund opportunities that are more limited for free users compared to their paid counterparts. This could be things like McKinsey providing broader access to deep research queries even after a free user has hit their limit. Where in this case you're not being served a click through offer from McKinsey but instead a version of the deep research experience that is presented by McKinsey and branded as such. Think training mode presented by Nike where your experience around a particular set of athletic goals triggers a branded experience within ChatGPT that again is presented by Nike. There's so many interesting things you could do with these premium brand formats. You could even have brands fun features that wouldn't exist otherwise. There is lots and lots of opportunity to create premium branded experiences where users aren't just being served transactional ads. Let's take it a step farther though and create branded action agents where the ad itself is a product, not a placement. Now OpenAI is already starting to walk down this path with their apps, but this would be the next obvious level where brands build constrained mini apps inside ChatGPT. Users explicitly opt in to that branded environment, use it for a discrete purpose and exit. When done, the UI could expand to encompass the brand. Some of the obvious examples that come to mind are the American Express Travel concierge or the TurboTax tax prep assistant. Now admittedly ChatGPT is already walking down this path and I think it's an extremely promising area because of all the different ideas for ad units, this is one where they're likely to be able to prove value to users by actually doing something useful for them in context right away. And so I think it would behoove them to put a ton of energy into this particular area. Lastly, what if we try to actually create ads that people root for? In their announcement post, OpenAI talked about how ads can be a level playing field for small businesses and new companies, but let's create a grants program that is explicitly for that. Founders Grants ad credits for businesses built with AI tools give one person unicorns the distribution leverage they lack. Small business grants ads that actually help small businesses reach broader audiences, create a Kickstarter type of energy and once again don't just run ads silently, but also show the recipients let users browse. So here we have an example of InterviewCoach AI where the ad unit itself shows you that it's part of the AI Founders Grant program and when you click on that you can go browse other recipients from categories like AI Built startups, local small businesses, or creator businesses. There are of course still people who are going to be cynical and just never care about advertising but in this world of AI there are going to be so many new types of businesses that are built that I think that there is lots of opportunity to tell the story of this next generation of businesses through the medium of advertising as well. Ultimately are most people going to care about any of this? The answer is no. Ads are to some extent just an inevitability and a necessary evil. But if I'm OpenAI even if I know that to be the case I would not simply be content to just make ads as unobtrusive and clearly labeled as possible. I would set the goal an ambitious goal to be sure of making ads that are actually value additive. Maybe you don't hit the mark but I think there's value in the attempt. Anyways guys that is going to do it for today's holiday episode. Tomorrow we will be back with our normal format. For now appreciate you listening or watching as always and until next time peace.

0:00