Perplexity Abandons AI Advertising Strategy Over Trust Worries

[MacRumors]

AI company Perplexity is stepping away from advertising over concerns that it will erode user trust, despite moves by rivals to introduce ads as an alternative money-making strategy.

Perplexity was one of the first AI services to embrace ads in 2024, after it ran tests where sponsored answers appeared under the chatbot's answers. That approach however was phased out last year, and executives at the company now say they don't plan to revisit it, according to the Financial Times.

"A user needs to believe this is the best possible answer, to keep using the product and be willing to pay for it," a Perplexity executive told the publication.

The report follows OpenAI's move earlier this month to show ads to ChatGPT users who have a free account or a low-cost Go subscription. OpenAI has said ads will not influence the answers that ChatGPT provides, nor will it provide advertisers with content from ChatGPT conversations.

Anthropic, the makers of Claude, recently mocked OpenAI for its decision to show ads to users and has said it has no plans to do the same. The company argues that including ads in Claude would not be in line with its mission of creating a helpful assistant for work and deep thinking, and that users should not need to second-guess whether an AI is being helpful or "subtly steering the conversation towards something monetizable."

Google features advertising in AI mode and in its AI Overviews summaries on traditional search results. However, Google has not introduced ads into its Gemini chatbot so far.

Ad strategies are one way that AI companies have been looking at as a way to generate revenue from users and reassure investors while spending heavily to train and operate large language models. Meanwhile, the cost of training and running large language models continues to climb, with no profit to show for it.

Article Link: Perplexity Abandons AI Advertising Strategy Over Trust Worries

 

[DBZmusicboy01]

The A.I. bubble is going to pop at any moment now 😆 most companies are figuring out they have created a monster without any regulation/safety switch.

 

[boswald]

The A.I. bubble is going to pop at any moment now 😆 most companies are figuring out they have created a monster without any regulation/safety switch.

And it can't happen soon enough.

 

[wikiverse]

The A.I. bubble is going to pop at any moment now 😆 most companies are figuring out they have created a monster without any regulation/safety switch.

A stupid monster. ChatGPT can't remember basic things that have already been established within the same chat thread, and I've tested them with some real-world tests for humans, given the same guidelines and questions and they only get about 40% correct.

The problem with monetizing LLMs is that they're not that useful. They're basically only good at summarizing things you've given them, and even then they regularly misinterpret, hallucinate or remove critical information. By the time you've double and triple checked their work, and corrected it, you could have just done it all yourself.

They're just not worth paying for at the moment. Their video and photo generation is getting better, but their basic logic, reasoning and understanding is not improving at all, in some cases I think it might be getting worse.

 

[cjsuk]

A stupid monster. ChatGPT can't remember basic things that have already been established within the same chat thread, and I've tested them with some real-world tests for humans, given the same guidelines and questions and they only get about 40% correct.

The problem with monetizing LLMs is that they're not that useful. They're basically only good at summarizing things you've given them, and even then the regularly misinterpret, hallucinate or remove critical information. By the time you've double and triple checked their work, and corrected it, you could have just done it all yourself.

They're just not worth paying for at the moment. Their video and photo generation is getting better, but their basic logic, reasoning and understanding is not improving at all, in some cases I think it might be getting worse.

Exactly that.

I note that you say that they're good at summarising stuff then immediately discredit it, so I'd argue they are no use for that too.

Anyway it gets worse as this post describes well...

People are tuned to the science fiction definition of AI when in fact we have LLMs which are statistical ******** generators.

The only reason they are as successful as they are is due to the relatively high level of incompetence of people who use them and the marketing job.

 

[Will Co]

The A.I. bubble is going to pop at any moment now 😆 most companies are figuring out they have created a monster without any regulation/safety switch.

...that they're going to struggle monetise....

 

[HouseLannister]

The A.I. bubble is going to pop at any moment now

"Pop" is just another word for consolidation. When this bubble "pops", small players like Perplexity will be acquired by bigger players like Google or Microsoft. AI doesn't just disappear. It just ends up controlled by a smaller group of companies with more resources.

 

[Jarman74]

Wow. In the AI world, tides seem to turn fast.

Just a few days ago OpenAI looked poised to pave the way for ads across the industry, and today it might find itself forced to reverse course just as quickly, suddenly becoming the odd one out.

 

[macmyworld]

Perplexity did not say how they will turn a profit. Most likely never.

The money spigot is slowing as hype does not turn into reality. AI is fun when free. Good luck getting enough people to pay for it.

 

[lepetitnicolas]

"A user needs to believe this is the best possible answer, to keep using the product and be willing to pay for it" a Perplexity executive told the publication.

Exactly !!! Put the ads where they belong : to the trash. Make a product, sell it (ideally without a subscription). If people want to try it before buying it then offer a X days "try it for free" then if you like it you buy it if you don't you uninstall it/delete your account. If the product is great it will sell, people will give money for it. If the product is crap it won't sell.... wants to make money ? Then make a great product. It's great but don't make enough money ? Then try a new market, change something, make it connect with people, make another product, something new and greater, follow your creativity, dream of something impossible and make it but for God sake don't ever include ads in it !!!

 

[Mac Fly (film)]

I hear eliminating whistleblowers and ignoring all concerns of their parents around the event can affect trust.

 

[Falco McGregor]

Remember when Wall Street analyst and sartorial nightmare Dan Ives said Apple will fix its AI problems by buying Perplexity? This despite Perplexity not actually making the foundation models that Apple desperately needs to compete in the AI race. Most of the people investing in the AI arms race have are gambling that this will be the biggest paradigm shift in the last 30 years if not more, which is causing a lot of skepticism and critical thinking to go straight out the window.

 

[contacos]

A stupid monster. ChatGPT can't remember basic things that have already been established within the same chat thread, and I've tested them with some real-world tests for humans, given the same guidelines and questions and they only get about 40% correct.

The problem with monetizing LLMs is that they're not that useful. They're basically only good at summarizing things you've given them, and even then the regularly misinterpret, hallucinate or remove critical information. By the time you've double and triple checked their work, and corrected it, you could have just done it all yourself.

They're just not worth paying for at the moment. Their video and photo generation is getting better, but their basic logic, reasoning and understanding is not improving at all, in some cases I think it might be getting worse.

What was truly eye opening to me was when I asked it to spoiler me on some tv shows I am currently watching (NOT recent shows) and it literally wrote plot points out of its ass but sounded so convincing. When I told it, that this didn't actually happen on the show it went like "ooops you are right. This did not actually happen. ..."

If you can't trust it for basic thing like that, well

 

[onepotatoetoo]

Perplexity did not say how they will turn a profit. Most likely never.

It's perplexing.

 

[neotint]

Perplexity’s start: we’re the anti-Google we’ll never be ad based, that is what ruined search and the Web.

Perplexity lacking true PMF with nerfed VAR: we love ads, they are the future of stochastic-based search I mean “answer engines”

Perplexity still lacking PMF with nerfed VAR: no, really what we meant before, ads bad.

So many Silly Con startups lack ethical leadership. The world needs to see SV culture for what it is: shiny marketing inflated pump and dump machines that con many. Really hope Apple leadership builds on its skepticism and leap frogs this whole stochastic unsmart path.

 

[WarmWinterHat]

Perplexity did not say how they will turn a profit. Most likely never.

The money spigot is slowing as hype does not turn into reality. AI is fun when free. Good luck getting enough people to pay for it.

Have any of them?

I know plenty of people that use AI, but no one that pays for it.

 

[iMac The Knife]

The A.I. bubble is going to pop at any moment now 😆 most companies are figuring out they have created a monster without any regulation/safety switch.

I've been hearing this for a few years now...

 

[iMac The Knife]

Have any of them?

I know plenty of people that use AI, but no one that pays for it.

Paying for AI today is like paying for a Vision Pro headset. You're paying for something that is slightly better than useless in its current state. Someday, both AI and AR/VR will be amazing, but today is not that day.

 

[CarAnalogy]

A stupid monster. ChatGPT can't remember basic things that have already been established within the same chat thread, and I've tested them with some real-world tests for humans, given the same guidelines and questions and they only get about 40% correct.

The problem with monetizing LLMs is that they're not that useful. They're basically only good at summarizing things you've given them, and even then the regularly misinterpret, hallucinate or remove critical information. By the time you've double and triple checked their work, and corrected it, you could have just done it all yourself.

They're just not worth paying for at the moment. Their video and photo generation is getting better, but their basic logic, reasoning and understanding is not improving at all, in some cases I think it might be getting worse.

It is getting worse. ChatGPT used to be a lot better but now it talks to me like a Gen Z or whatever these kids are and is constantly putting on unnecessary guardrails.

I swear if it says vibe to me one more time.

 

[nt5672]

I try to use various AI tools every couple of months thinking they must be getting better, but no they are not.

 

[8KYUP]

Exactly that.

I note that you say that they're good at summarising stuff then immediately discredit it, so I'd argue they are no use for that too.

Anyway it gets worse as this post describes well...


View attachment 2606057

People are tuned to the science fiction definition of AI when in fact we have LLMs which are statistical ******** generators.

The only reason they are as successful as they are is due to the relatively high level of incompetence of people who use them and the marketing job.

 

[8KYUP]

Surprised by the number of people here on a tech forum that actually think AI is going away. There has never been a technology that has gone from ~ unused to essential in so short period for so many users. That has occurred in a single year, even for the technologically uninclined. Hoping it away is simply silly, head in the sand behavior.

Are the security concerns real, absolutely. Are the market bubble concerns real? Likely, and to what degree we'll find out. Is AI going away? Absolutely not.

Companies like Anthropic see less than 20% of their revenue from individual users who would potentially be looking at advertising. Most is enterprise. Google/deep mind usage of their AI is interwoven into all of their products and is not directly reliant on consumer spend for the lion share of what they're doing. OpenAI is an outlier there as they are largely consumer based right now. Focusing on monetizing that one segment is an interesting sidenote but far from the big picture.

 

[Rix1010]

I was actually wondering this morning what I would do if AI ceased to exist. And I realised not that much would change. I’d just go back to searching for information on Google instead of asking a chat bot.

This clearly doesn’t bode well for Altman and his ilk. When you consider their diatribes on AI curing cancer and AI replacing most jobs, you have to wonder what they were thinking.

I barely use AI for professional or personal reasons, and I used to be fairly pro-AI a few years ago. As aforementioned, using AI often slows me down.

I’m slightly concerned as to what the AI bust will look like—and hope that not too many everyday investors get wiped out. At least, if the bubble does pop, memory prices will come down. (Swings and roundabouts).

 

[Apple_Robert]

The company replies should say, ‘we will not place ads in our app / service. Instead, readers are given the ‘“we have no plans to…” There is a large chasm between those two positions and one that users of AI need to keep in mind.

 

[WarmWinterHat]

Surprised by the number of people here on a tech forum that actually think AI is going away. There has never been a technology that has gone from ~ unused to essential in so short period for so many users. That has occurred in a single year, even for the technologically uninclined. Hoping it away is simply silly, head in the sand behavior.

Are the security concerns real, absolutely. Are the market bubble concerns real? Likely, and to what degree we'll find out. Is AI going away? Absolutely not.

Companies like Anthropic see less than 20% of their revenue from individual users who would potentially be looking at advertising. Most is enterprise. Google/deep mind usage of their AI is interwoven into all of their products and is not directly reliant on consumer spend for the lion share of what they're doing. OpenAI is an outlier there as they are largely consumer based right now. Focusing on monetizing that one segment is an interesting sidenote but far from the big picture.

I don't see anything saying it's going away, just that it's a bubble. Dot.com was a bubble, and the web didn't go away...and I don't see any industry, outside of tech, where AI is "essential". Not everyone works in tech; I don't. Just a guess, obviously, but I think around 10% of the current crop of AI companies will still be around in 5 years, and the survivors likely won't be pure AI providers, like Google, for example.

Now if someone can come with an equivalent usable system that doesn't require the power of a thousand suns, and enough hardware to build death star, they will be the winner....The current direction of LLMs is unsustainable for a multitude of reasons, and the quicker it dies, the better.