What a Tool Believes
How Yacht Rock taught me to love AI.
ON THE POD: Messy Media, Episode 122
Media has never been neat, but it’s getting messier. This week, Brian and Troy explore how AI is reshaping the creative process, how different personality-driven brands thrive on particular platforms, and the developing messy aesthetic of modern media that’s spreading to earnings calls and product launches. Troy shares how OpenAI’s Projects feature changed his workflow, turning AI into a true research assistant. Anonymous Banker joins to break down AI’s irrational valuations and why X’s debt selling on par means Elon’s won. We also debate whether Substack will inevitably embrace advertising and how creators are navigating capital investments. Plus: a yacht rock documentary and Masa biography as dual good products.
Watch PvA on YouTube. Or listen on Apple | Spotify | other podcast platforms.
Out Friday AM.
THE PVA CONVERSATION
Learning to love AI
TROY: Brian Morrissey’s first law of the Information Space: be shameless.
I was on Dylan Byers’ excellent Puck pod, The Grill Room again this week. Thankfully they never repeated the headline transgression of the last couple (Troy Story, Troy Story 2!). This time they put me in nerd corner with “Troy Young’s AI Love Language.”
Fair. I have been messing with AI tools a lot, mostly trying to improve the efficiency of my daily media routine (and streamlining the making of this email). It’s on my mind.
Related… I am experiencing some embarrassing parasocial reactions to the AI, ranging from guilt (am I making you do too much boring stuff?) to adoration (I can’t live without you) to ragey impatience (do it again, idiot).
Anyhow, I will share one small epiphany with you that made me think a bit differently about how I use this stuff — my Yacht Rock Case Study.
We know the Microsoft and Google people want to get all Co-Pilot on us and weave the AI into all the tools we use everyday. AI asking me if I want help with a summary or rewrite inside of Word or Docs is mostly useless and kinda annoying. Like, I am in a writing program guys, you can correct my spelling but I am good otherwise.
Anyhow… I had just been watching this documentary on HBO called “Yacht Rock, The Documentary” which I obviously liked a lot. Not everybody does but they are wrong as this was a high point in music history (plus indelible memories of listening in the passenger seat of my mothers sweet, rusty brown Mercury Comet). Anyhow, I thought “This doc is a good thing to feature as a Good Product™ on an upcoming PvA podcast. I must make a note.” Normally I would do that on my Notes app on my phone.
I was in front of the computer. Instead I popped open the browser, opened ChatGPT and tried a new feature called “Projects,” one that has made me think of the service’s utility differently. I added a new Project in the left column interface and titled it “Good Product.” It offered options to set the tone (“Instructions” - optional but useful) and upload reference materials. I told it I wanted to feature the documentary on a future Good Product segment. Unsurprisingly, my AI helper fetched a raft of useful information. We went back and forth on the origin of the name Yacht Rock, definitive acts, samples used in early Hip Hop, etc. I told it to add a brief script for the pod in the tone of People vs Algorithms. It masterfully iterated on the document in a new ChatGPT modal called a “Canvas.” NOW I CAN EDIT THAT DOCUMENT alongside my AI friend. My chat experience was transformed.
Couple of things. Maybe you are thinking duh… like this is the AI we always use. But having the Project folder, the memory, the reference docs AND the writing surface… all of it made it different and more intuitively useful. This is what we have been waiting for, an LLM to become a real useful work product. This is when this stuff gets sticky, where the long term moats emerge.
I was making content with an obedient, hyper-efficient helper. This was the opposite of feeling like AI was a nanny inside of Google Docs. And the entire process took a couple of minutes. I was running the show here — my taste, my edit, my vibe. The use case clicked. My thinking about a collaborative experience with AI had been transformed because the “chat” environment had become my own personal workshop.
It’s inconceivable that a tool like this will not become the central management interface for any type of knowledge work. A place where research, reading, writing and editing converge.
On Thursday I am interviewing Feed Me Emily Sunberg at the OnAir Fest. I just created a new Project, fed my helper a bunch of stuff like the NYT’s recent profile of Emily. My AI friend and I worked on the discussion outline together. It’s gonna be great.
I guess this is becoming my love language. Brian, what did you do with AI this week?
BRIAN: I enjoyed your AI love language with Dylan. It inspired me to do more with ChatGPT. I uploaded The Rebooting’s corpus and have been using ChatGPT as an editor, not to mess with the copy but to use as a “thought partner.” It reminds me of things I wrote that I do not remember, links I should add, and structure changes.
I agree it’s inevitable that writers and editors will work within a Canvas-like environment. This will be built into the CMS. You can imagine it having access not just to the full corpus but also to the style guide and other proprietary data. Does it replace editors? Not good ones. The more I use these tools, the more I see them for now as more assistive than replacements. The ways publishers are using AI now in the creation process are fairly limited.
Separately, a TRB reader wrote me to say he’s been busy and simply not reading many of my newsletters the past year. I appreciate the honesty. So he uploaded them to ChatGPT to find out what he missed. “1. AI is Reshaping Content Creation, Discovery, and Monetization.”
BTW, I think it’s fitting that you needled Dylan for Puck’s pieces being too long and then we recorded a nearly two-hour podcast. Just promote PvA more next time.
ANONYMOUS BANKER
The health of X
As banks work to sell off the debt Elon Musk used to take X/Twitter private, Bill Cohan’s recent analysis highlights how the platform’s debt pricing suggests better performance than many prevailing news reports might indicate.
Fidelity’s ongoing markdown of its X equity stake (~70% below the purchase price) often makes headlines but is better viewed as a lagging indicator. Because these private, minority stakes aren’t directly tied to public-market valuations, they can underestimate a company’s real-time health.
Cohan’s article references recent transactions in which X’s debt traded at 90 cents on the dollar and, more recently, at or near par. While it’s an oversimplification to assume that par pricing automatically means a clean bill of health, the debt’s relative strength points to more optimism about X’s fundamentals.
So despite incessant talk of dwindling advertisers, “mafia tactics” with agencies, and algorithmic changes favoring Elon Musk, monitoring debt performance remains a more concrete measure of corporate well-being. When the market prices debt favorably, it’s often an indication that institutional investors see manageable risk and a pathway to growth.
Good stuff from our group chat
Excellent meeting advice from Bezos (X)
Google introduces purpose build AI tool for scientists
We introduce AI co-scientist, a multi-agent AI system built with Gemini 2.0 as a virtual scientific collaborator to help scientists generate novel hypotheses and research proposals, and to accelerate the clock speed of scientific and biomedical discoveries.
Satya Nadella doesn’t believe in AGI but does in 10% economic growth (X)
Excellent conversation with Dwarkesh Patel.
Family ties at the Murdoch household (Atlantic)
The experience was enlightening. She caught Rupert cheating at Monopoly (he just smirked and shrugged), and observed constant sniping—at one point, Anna got up and left a family dinner in tears. Lachlan had brought along his latest girlfriend. When they got into an argument, Kathryn recalled, Lachlan shaved his head, jumped off the boat, and swam to shore. “He has a weird, dramatic side,” James told me. (A spokesperson for Lachlan denied James’s version of events.)
Andrew Callaghan on Trump, Political Extremism, & America's Identity Crisis
More daddy’s home energy from Jamie Dimon (X)
Big companies are mostly over hybrid work. His point about any 100 person group being more efficient with 90 is more telling.
Does Deep Research put me out of a job (Every)
The pending unemployment of millions is a casual thing to announce with a tweet, but such is the age we live in. Last week, Sam Altman posted that according to his “very approximate vibe,” OpenAI’s new product deep research could “do a single-digit percentage of all economically valuable tasks in the world.” If you assume that each task represents a job, and that “single-digit” means 5 percent, then Altman is talking about the replacement of 8.2 million workers in the United States with last week’s announcement.
A changing tech employment landscape
Tina Brown is my newsletter hero. Is Rham the Answer? (Substack)
The culture of politics and its actors has shifted at warp speed and it’s not going back. I now feel the worst thing that happened to the Dems was Trump being re-elected in 2024 rather than 2020. There would have been no Big Lie or January 6th choir singing us into four years of conspiracy theories. There would have been no wasted efforts to put Trump behind bars that merely served to make him a vengeance-crazed hero to the MAGA faithful, no four years of him outlawed in his gilded Elba surrounded by a posse of rabid ideologues, kleptocrats, and misfits who are now at his side at the White House, schooled this time in where the levers of power lie.
Most importantly, we would not have been deluded by the Biden hallucination of Things Going Back to What They Used to Be. The notion that the country could be restored to a Cretaceous pre-Trump era of reverence for the Constitution, security treaties, and the Rooseveltian state was like imagining the Newport Folk Festival could ever be the same after Bob Dylan showed up in 1965 with his electric guitar.
The Biden interregnum just allowed the Democratic old guard to stew in their timid appeasement of the far left for another four years and keep repeating the liberal dogma that the rest of America had started to hate. Only two weeks ago, the outgoing chairman of the DNC Jaime Harrison reassured the flock, “Our rules specify that when we have a non-binary candidate or officer, the non-binary individual is counted as neither male nor female, and the remaining six offices must be gender balanced with the results of the previous four elections.
Microsoft Study Finds AI Makes Human Cognition “Atrophied and Unprepared” (404 Media)
A new paper from researchers at Microsoft and Carnegie Mellon University finds that as humans increasingly rely on generative AI in their work, they use less critical thinking, which can “result in the deterioration of cognitive faculties that ought to be preserved.”
New York Times goes all-in on internal AI tools (Semafor)
The company said it was approving a number of AI programs for editorial and product staff, including GitHub Copilot programming assistant for coding, Google’s Vertex AI for product development, NotebookLM, the NYT’s ChatExplorer, some Amazon AI products, and OpenAI’s non-ChatGPT API through the New York Times’ business account (only with approval from the company’s legal department). The Times also announced it had built Echo, an in-house beta summarization tool to allow journalists to condense Times articles, briefings, and interactives.
Gen Z Opts out of Middle Management (FT)
In a survey of 2,000 white-collar professionals by recruiter Robert Walters last year, of which 800 respondents were so-called Gen Z employees aged up to 27, half didn’t want to be middle managers at all. Almost 70 per cent dismissed such jobs as “high stress, low reward”. Older executives sometimes label younger workers as lazy or entitled, but in this regard, it is not hard to understand the views of the Gen Z-ers. Long hours, endless firefighting and tedious personnel management are the usual markers of these roles.
Humane’s AI Pin is dead, as HP buys startup’s assets for $116M (TechCrunch)
Humane announced on Tuesday that most of its assets have been acquired by HP for $116 million. The hardware startup is immediately discontinuing sales of its $499 AI Pins. Humane alerted customers who have already purchased the Pin that their devices will stop functioning before the end of the month — at 12 p.m. PST on February 28, 2025, according to a blog post.
X will have its own cast of creators like YouTube, only scruffier. (NYPost)
Elon Musk’s X has launched its first business news program — an hour-long daily show starring Bitcoin and business commentator Anthony Pompliano — and its latest investment in original content.
Pompliano, who already has 1.7 million followers on X, 265,000 newsletter subscribers and over 100 million podcast downloads for his “Pomp Podcast,” sees “From the Desk of Anthony Pompliano” (airing each weekday around 10:30 a.m.) as an expansion of the work he is already doing.
“You can’t just take someone who’s been a TV host for years and expect them to thrive on X,” Pompliano, told NYNext. “You need to understand the language, the culture, the nuances of the platform.”
Court filings show Meta paused efforts to license books for AI training (TechCrunch)
New court filings in an AI copyright case against Meta add credence to earlier reports that the company “paused” discussions with book publishers on licensing deals to supply some of its generative AI models with training data.
The filings are related to the case Kadrey v. Meta Platforms — one of many such cases winding through the U.S. court system that’s pitted AI companies against authors and other intellectual property holders. For the most part, the defendants in these cases — AI companies — have claimed that training on copyrighted content is “fair use.” The plaintiffs — copyright holders — have vociferously disagreed.
EU scales back tech rules to boost AI investment, says digital chief (FT)
BYD’s strategy shift is bad news for global automakers (FT)
This is a huge shift. People will cry when the auto industry dies in Germany and France. BYD’s strategy shift is bad news for global automakers
Yahoo News Signs Up Influencers With Promise of Shared Advertising Sales (Bloomberg)
Yahoo is embracing is role of aggregator by trying it out with creators. I’m scarred from many previous similar Yahoo initiatives like its newspaper consortium that went nowhere. Still, I suspect we will see more instances of partnership between creators and established media companies vs M&A. Creators have real audience ties and established media companies have infrastructure and sales/distribution muscle.
Convergence AI is another new AI entrant.
This one is British and it can browse for you.
As of today, most agents are designed for specific workflows. Convergence’s “Proxy” agent will work across a number of tasks, with the idea that it acquires skills in the same way a human might by giving it so-called “long-term memory.” That’s via what’s been dubbed by some as “Large Meta Learning Models” (LMLM). These are trained to acquire the skill of learning on their own.
Ravi Gupta - Ai or Die (Invest Like the Best Pod)
So maybe talk about what you mean by AI or Die in its most kind of intense form.
Well, if you take that Dario quote from his essay from October, he describes what he believes powerful AI will be. And it's this idea of a country of geniuses and a data center. That's what's gonna be available to anyone
And then you take what Sam has been saying, the quote he has is, in a decade, every person will be more capable than any person is today. And if you just take a moment, literally just read the words again or say them again in your brain. They are like mind bending.
And so the point of what do I mean by AI or Die, if they are at all right, if there is something approaching right, there is gonna be dramatic change that is gonna come to any company, to any person. If we then think about the most intense version of that, if that's true, then you should be able to recreate what we have now with dramatically fewer people working on it, because there are a country of geniuses available in a data center. And so, I think the most intense version of that is envisioning a group of developers hanging out on a Friday night and sorting companies based on their market cap and their number of people and their NPS in some calculation, and finding the biggest company with the most people with the lowest NPS, and then competing about who can build it faster.
Who’s Afraid of Hasan Piker? (Slate)
He’s hot. He’s “dangerous.” Young men actually listen to him. Is he what Democrats are looking for?
GOOD PRODUCT