Microsoft to invest $10B in OpenAI, generative AI hype, America's over-classification problem

The transcript opens with studio chatter that the automatic transcription renders poorly. One speaker asks whether anyone else sees a half-second to one-second lag on a colleague's video, joking that it comes from the way his mouth moves, and there are test phrases, a check that recording has started and a few lines of banter that are too garbled to interpret. Nothing substantive is said here, and fragments such as the line about open-sourcing something to the fans cannot be reliably read.

The hosts then welcome listeners to episode 111 and joke that it is the world's greatest podcast "according to Slate". A host adds that this is the podcast the press cannot mention, and says the outlet had just published a profile of the show.

One host claims the show is the number one business and tech podcast in the world and argues that this success leaves the press struggling to give it attention because journalists "want to hate us". Another host says the Slate writer's point was that the hosts avoid independent questions by going direct, a tactic that is now common. The hosts respond with irony: the real question, one says, is why the subjects of stories prefer not to have the press interpret what they say. These are the hosts' own characterizations and are not supported by any outside evidence in this passage.

A host concedes the downside: going direct means avoiding independent scrutiny of their views and decisions. Another replies that the press keeps writing "hit pieces" about them, so there is little reason to run their side through a filter that will produce one anyway. A third claims journalists have a class hatred of technology, entrepreneurs and investors.

The speaker says the press treats people better when they agree with its biases, and cites what "SPF did" as the example. The claim is the speaker's own opinion, and the transcript gives no detail on what that refers to.

When asked which specific large media outlets "they" means, a speaker offers a trade: you may name Fox, he says, but he can list MSNBC, CNN, the New York Times, the Washington Post and the Atlantic. Another host raises Megyn Kelly, and the reply is that she is now an independent podcaster rather than part of the mainstream. The speaker insists that no outlet falls outside this mainstream group and that they all think alike with only small differences.

The hosts attribute this to clicks and "advocacy journalism". A host objects that advocacy is just bias, and the response is that it is activism. One speaker says he coined the term advocacy journalism, a claim the transcript offers without support. Another host notes there are also journalists who do not fit this description, to which a speaker answers that it depends on the topic and the outlet.

The speaker's broader argument is that many journalists work for the same reason the hosts do, to gain influence, since they are not paid much, and that their route is pushing a political agenda. He calls them activists and "party actors". This is an opinion about motive and is presented without evidence.

One host points to Draymond's podcast as a sign of the trend: it has become a popular form of sports media in which he can speak without journalists' filtering and classification. The host says audiences want direct, unfiltered points of view, but that journalism may still have a separate role in scrutinizing, analyzing and questioning those voices. The answer from another host is that this is activism, not journalism.

Later a host presses the same point from the journalist's side. A principal tends to understand the material better than an outside analyst, but someone with direct access is also an agent with a point of view and a direction to push. So, on this account, journalists can in theory serve as off-field analysts. The reply from a fellow host is that journalists are less educated and more biased than the hosts, a subjective claim that is not backed up.

A host brings up an article about the Fed and the debt ceiling in which, he says, the writer did not understand the difference between a percentage point and a basis point. The name of the writer is garbled in the transcript, and the claim about the article's contents is the host's description rather than something that can be checked here. Another host recalls seeing it, and the group reacts with disbelief at the idea that the Fed was raising rates by 25 percent rather than 25 basis points.

Another host calls the point undeniable on one axis: whatever one thinks of who is more educated or biased, there are zero checks and balances if such a mistake survives proofreading and editors. His inference is that other, less obvious errors must also get through, and that the average person on Twitter cannot police all of it. He suggests unforced errors made in the search for clicks could tip readers toward one conclusion over another, which is his speculation.

A host proposes a simple test: when you read a journalist on a topic you know well, you see they grasp only some 10, 20 or 30 percent of what is going on, yet on a topic you do not know, such as Hollywood or an unfamiliar region, you assume the story is fully correct. Another speaker names this the "gel man amnesia effect" in the transcript and says it came from Michael Creighton. The percentages are an illustration, not measured data.

The first host says the problem is worse now because mistakes come not only from sloppiness or lack of expertise but from an agenda. As evidence he points to the Slate piece itself: he says the article was not bad and made the hosts look cool, with a subheadline about a "safe space" for "Silicon Valley's money men". The headline, he says, changed to say Elon Musk's inner circle is telling us exactly what it thinks.

The hosts say this headline is a bid for clicks that frames the show through Elon, though Musk figures in perhaps two of 110 episodes by the host's count. The same host says an earlier headline word was "cronies" and that someone edited it to "inner circle", and that he saw the earlier wording in a tweet summary. This is a secondhand recollection from one host and the transcript shows no verification.

Continuing the Slate complaint, a host says the earlier, harsher headline word was frozen in time in social media summaries after someone reconsidered it and toned it down. He then describes what he says happens in the editorial process: an article may or may not be edited or proofread, and in a race for traffic a social-media specialist picks six or seven headlines and A/B tests them, sometimes with software, and sometimes with a paid test of about $5 in ads. The best performer wins. This is the speaker's account of industry practice, offered without sources.

His conclusion is that this is cynical because many readers see only the headline and take it as the story. He tells another host that after a New Republic piece with a bad illustration, the advice was not to worry, since people just read the headline and few read the story.

The subject of that profile says the piece itself was well written and that he was shocked a writer had taken the time to produce decent prose. Another host adds that the writer had clearly listened to many episodes. The group recalls the earlier advice to look only at the picture and move on if you are fine with it. One host says he doubted that at first, but it turned out to be mostly true. Jokes follow about unflattering pictures, including a remark that another person looks like Hugh Grant in Notting Hill.