MediaMorph
Posts
MediaMorph Edition 65 - by HANA News

MediaMorph Edition 65 - by HANA News

Perplexity becomes even more perplexing

Mark Riley
August 19, 2025

In partnership with

Brought to you by Mathison.ai

MediaMorph Edition 65 - by HANA News

Perplexity becomes even more perplexing

Was this newsletter forwarded to you? Sign up here

The written-by-a-human bit

There was an illuminating moment when Mark Zuckerberg and his entourage came for a visit to my office in NYC about 8 years ago to visit News Corp, WSJ and Fox News. While being shown around, Zuckerberg was genuinely surprised to learn that the news division and the opinion section were independent entities with separate reporting lines.

At about the same time, Zuckerberg’s Harvard dorm mate and fellow Facebook founder, Chris Hughes, was struggling to contain an editorial revolt at The New Republic, where his West Coast Silicon Valley approach had led to a full-blown crisis.

I mention these to flag an ongoing and persistent mismatch between tech and media cultures. The allure of progress and convenience, plus the seduction of VC money and power, crash up against the fragility of truth and trust.

Again and again, legacy media mastheads are at the mercy of the tech barons who own and direct the digital traffic on which they depend. For editors and proprietors, the terrifying question is “do they understand us, do they care?”

Which brings us on to Aravind Srinivas, CEO of Perplexity AI, a 31-year-old Indian computer scientist who recently made a paper bid to buy Chrome for $34.5 billion (roughly double the current value of Perplexity). Cynics might suggest this is simply a PR stunt, designed to boost Perplexity’s already ballooning brand. But Perplexity has already been accused of hoovering up and caching paywalled content to power its popular answer engine. Which begs the question, does the scientist Srinivas understand or care about the creatives and content creators?

This was a question put to him by Casey Newton and Kevin Roose on NYT’s Hardfork podcast last week.

His answer was perplexing:

“Well, here's the thing.

There are two aspects here. One is like you're talking about the creators. No, there are like two types of creators. People who are actually really good. Like, for example, when you guys write something, people care.

And then there's lots of spammers and, you know, hacksters who just write like erroneous blog posts, erroneous content, like fake information, clickbait articles. I don't think it actually empowers the user, right? You're only talking about the creator, but you have to consider the user as well.

And so for the first time, AI is in the hands of users through like agents that actually go and do stuff for them that take into account their instructions and protect them from all the spam. So we want to figure out a model that works for the user and the creators together. And penalize the bad creators and incentivize the good creators to just focus on wisdom and knowledge and truth and interesting stuff.

And by the way, even in a world where agents are doing all this stuff for people, the humans are still going to continue browsing the web. There are people who believe the web is going to be completely agentic. You don't even need a browser. Browsers are so 1990s. I don't believe that. If we believe that, we would never even launch a browser. We would just continue with the chat UI. So we believe people are still going to be browsing and surfing interesting things on the web. But we think that you should give users the power to decide how they want to do it. And first time have an AI that can protect them against spam and hacks. Now, how to monetize this, how to give the creators the right incentives here? We are going to announce something to that effect where publishers can be incentivized for creating interesting, good content. We think about it in two ends of the spectrum. One is like completely human-centric like Apple News, which is a pretty good model. And the other is just buying the content and training your models, like the licensing deals that Open AI has done with the Wall Street Journal. I think you want to be somewhere in between where you do want to say, okay, there's going to be some elements of AI here. It's not just going to be humans. And so you don't want to just build an Apple News-like model, but it's going to be closer to Apple News, with some protections for users to say they can have AIs also read those articles. And the publishers get rewarded. So that's how I'm thinking about it…

…My first point, which is basically that if you can delegate the boring things, the things that you don't want to be doing to the AI, you're just going to spend time surfing on things you actually want to be doing, actually want to be reading.

And then that puts the incentive on the creator to actually create really interesting, high-quality stuff. You can even charge even higher because people have more time. So if they're going to come to you, they're coming to you out of their own will. So they'll be willing to pay for it even more.

Now, there are a lot of unknown unknowns here on how it's actually going to roll out. But my belief is that the ones who have built a reputation and a brand for saying correct things... that stand the test of time are going to be able to charge even more for their content.”

So, stand by for an announcement about an AI-powered aggregator that handles the boring stuff and weeds out the spam, allowing us humans to browse for and pay even more for the more interesting stuff. Perplexity to the rescue! Sounds well-intended, and at least Srinivas is thinking about it — but so many questions about how this will happen, and given his track record, does he really understand or really care?

Mark Riley, CEO Mathison AI

At Mathison AI we are firm advocates for media companies building their own subscriber facing chatbots for deeper engagement.

The FT, Aftonbladet, Forbes, The Washington Post and Rappler are all leading the way.

We understand the need for RAG based training, guardrails and guidelines. We build smart AI that has the inbuilt humility to admit when it doesn’t know the answer.

Speak to us today about testing your archive in a safe enviroment.

Book a call today

Mark Riley [email protected]

AI and Journalism

I Recently Graduated With a Degree in Journalism— Here's How I'm Navigating My Career in the Age of AI

Success -

Recent journalism graduate Sarah Thompson is embracing the impact of artificial intelligence on media, aiming to specialise in data journalism by mastering AI tools for analysis and visualisation. She seeks internships at innovative media organisations to enhance her storytelling skills while blending traditional journalism with cutting-edge technology.

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

Join the Superhuman AI newsletter – read by 1M+ people at top companies
Master AI tools, tutorials, and news in just 3 minutes a day
Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

AI and Academic Publishing

Times Higher Education (THE)

AI-based fake papers are a new threat to academic publishing

Times Higher Education (THE) - August 15, 2025

As an associate editor, I've observed a troubling rise in AI-generated manuscripts that mimic existing research while evading plagiarism detection, often submitted by authors using fake affiliations and emails. This trend poses significant risks to scientific integrity, prompting the need for stronger safeguards, such as AI-authorship detection tools and enhanced training for editorial teams to recognise these deceptive submissions.

Academic publishers and AI – one year on

The Bookseller -

The ongoing debate in the publishing industry highlights the urgent need for fairer income distribution as authors face declining earnings amid the rise of generative AI. The Society of Authors advocates for equitable contracts, urging authors to seek tailored advice and negotiate better terms in light of publishers increasingly viewing them as mere content providers.

MediaMorph Edition 65 - by HANA News

Perplexity becomes even more perplexing

MediaMorph Edition 65 - by HANA News

Perplexity becomes even more perplexing

The written-by-a-human bit

Find out why 1M+ professionals read Superhuman AI daily.

Standard