316 - Sources of Credibility

Credibility, and discredibility, are contagious. Anyone sharing hype and marketing materials from OpenAI, Anthropic, and others are infecting themselves with discredibility at every occurrence. This is acutely true of "news" sources, whose explicit purpose is to reduce the bias of reporting to the minimum levels possible. In some cases this turns the name of that news source directly into an oxymoron, such as a recent article from "(un)Scientific American".

If a news source or "influencer" routinely circulates hype and benchmarks that are quickly debunked, they can be termed a "Disinformation Broker", and everyone citing them becomes a peddler of the same disinformation, spreading the contagion of discredibility far and wide.

Like any contagion, isolation from the infected is critical, combined with awareness of the problem, and a path which those seeking recovery may follow. Strict criteria must be upheld, either by an individual or their trusted intermediary (such as a platform, like what "news" is supposed to offer, in theory more than practice), dividing those who persistently distribute disinformation to the masses from those who invest effort in maintaining baseline credibility.

Among those with baseline credibility there are several common archetypes. Those who make themselves true "third-parties", keeping arms' length distance from everyone in efforts to distribute minimally biased information, embody what news is supposed to be. Those who work directly within domains can have both vested interests, and maintain their own credibility through disclosure and transparency, but only so long as they avoid the pitfalls of PR and hype.

Ultimately, anyone both working in and passionate about a domain is likely to have some vested interests, and some of their own efforts to solve some subset of problems therein. It isn't a mark against their credibility to have these interests, rather it can be a mark of consistency with their stated objectives. The opposite is also true, that if someone (outside of news) routinely talks about a domain, but they make no effort to solve any problems therein, that can be a severe mark against their credibility, as their words and actions are inconsistent.

Virtually nothing is more biased and full of unhinged disinformation about AI than "GenAI" itself. Ask any LLM or "agentic" derivative thereof about AI and it will regurgitate the same PR material, hype, and other BS that it was trained on and further biased with via secondary systems of explicit bias injection, like RLHF.

If someone or some "news source" you know is citing the PR material and blatant hype (not debunking) of OpenAI, Anthropic, or other similar companies, or they're quoting known bad actors like Geoffrey Hinton, Max Tegmark, Jan Tullin, Linas Beliūnas, Ethan Mollick, Eric Schmidt, or Allie K. Miller, cut them off, give no second chances, and have no regrets.