Have you ever thought that ChatGPT is too good to be free? That’s why OpenAI is trying to make money from its free users, with the most advanced models reserved for paying users. Like? With a highly tested, effective system that every Internet user is accustomed to: advertising. In fact, OpenAI is testing adding ads to ChatGPT.

Remarks. It was OpenAI CFO Sarah Friar who confirmed this information in statements to the Financial Times (via Axios). The executive said that they are considering adding ads to their chatbots, but want to “think about when and where to implement them.” However, after this statement was published, OpenAI published a statement that was also published in the American media. The statement included the following statements:

“Our existing business is experiencing rapid growth and we see significant opportunities within our current business model. While we are open to exploring other revenue streams in the future, we have no active plans to continue this growth.” [inclusión de] advert.”

So right now, advertising isn’t in the company’s immediate plans, but it seems like it’s an issue on the table.

Costs. Artificial intelligence is very good and really useful. Moreover, it is free. Anyone can start ChatGPT and solve a question by accessing the model of a $157 billion company without paying any fees. However, behind this is a process that requires a lot of power and consumes energy. Until now, companies have been generating revenue through payment models and private investments, but it was only a matter of time before this happened.

Example already. Perplexity, one of the most popular search engines based on artificial intelligence, also has this feature. Just a few weeks ago, the company announced that it would begin advertising trials in the United States, thus kicking off the season. For now its offering is based on sponsored suggested searches and is marked as such, but the truth remains: a chatbot advertising could raise more than one eyebrow.

This is what ads look like on Perplexity | Image: Astonishment

Be careful. The development of chatbots has targeted the pursuit of precision, meaning that artificial intelligence has the ability to respond in the most accurate way possible. There’s undoubtedly still room for improvement, but that’s the real focus. The other one is multimodal artificial intelligence, but it does not concern us today. The truth is that companies like OpenAI, Claude, Google, Microsoft, and Company need to be careful about how they implement advertising in chatbots and be very discerning about what is advertising and what is not.

The risk is clear. If a user asks what to look for when buying a bike and the chatbot responds with “let that model from that brand” because the brand in question paid to run that ad, the chatbot’s response will not be completely reliable. It will be conditioned by advertising and thus the user’s trust in this technology will decrease. Not to mention the paradox of the chatbot recommending a product and at the same time a similar product appears in an ad.

Advertising in itself is not bad.. As long as it is well differentiated and does not impact the AI’s results, its inclusion should not be too different from the current model we see with Google, Bing and other search engines. But the key is this; Draw a line, define very well what this ad will be and do not implement it in a way that can be easily confused with organic content.

OpenAI needs revenue. Actually many. Despite being a company valued at more than $150 billion and able to raise $6 billion 600 million in an investment round, the company maintains an expensive service and, as we see, is taking in more and more. It is estimated that OpenAI has annual losses worth $8 billion and has not yet managed to equalize expenses and revenue. If your goal is to be the new Google, which it seems like it is, the money needs to start flowing. If there’s one thing Google has taught us, it’s that advertising works.

Cover image | Matheus Bertelli

in Xataka | 230 million Chinese currently use chatbots. ChatGPT is one of them, even though it shouldn’t be there