Josh Bernoff, een voormalig SVP bij Forrester, is een interessante figuur in de Amerikaanse blogosphere. Hij werd breed bekend na het verschijnen van het boek Groundswell: Winning in a World Transformed by Social Technologies, waar hij de auteur was samen met Charlene Li (een ander Forrester zwaargewicht). Bernoff schrijft in een voor de Amerikaanse cultuur ongekend transparante en objectief manier, en is daarmee een belangrijke respondent van gezond verstand, een eigenschap die de VS in hoog tempo lijkt te verloochenen.
Vandaag wil ik grote delen van één van zijn meest recente blog postings onder de aandacht brengen; het betreft het zeer actuele onderwerp van zakelijke piraterij. Zie ook: Disruptor disrupt (beschrijving van zakelijke piraterij) en Pirates hate piracy (voorbeelden van zakelijke piraterij)
Nobody whines louder than a disrupted disruptor
Chinese AI model Deepseek has thundered onto the scene, competing effectively with existing players like OpenAI and Meta while using far less computing power. Global markets reacted immediately, knocking half a trillion dollars off the market cap of Nvidia, whose chips power most of the big AI models. Short sellers of Nvidia made billions.
OpenAI execs reacted immediately by — of course — accusing Deepseek of cheating. As reported in the Wall Street Journal:
“It is critically important that we are working closely with the U.S. government to best protect the most capable models from efforts by adversaries and competitors to take U.S. technology,” an OpenAI spokeswoman said. China-based companies are “constantly trying to distill the models of leading U.S. AI companies,” they added.
What is distillation? It’s a method of training smaller AI models on the output of larger, more extensive, more costly models. And according to OpenAI, it’s illegal. David Sacks, Donald Trump’s AI czar, claimed that, “there’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models . . . And I don’t think OpenAI is very happy about this.”
Let’s unwind what’s happened here. Large Language Models (LLMs) exist because they hoover up massive amounts of web content to learn from. The highest quality content is in books — which is why many of them were trained on pirated book content. But most of the content they swallow is just ordinary copyrighted content available on the open web. It’s an open question whether training on copyright content is legal. That’s at the heart of copyright suits from organizations like the New York Times against OpenAI.
So OpenAI’s argument is basically “It’s fine if we steal from copyright owners, but you’re cheating if you steal from us.”
Bernoff stelt verder dat dit de standaard modus operandi is van Silicon Valley:
This twist is ironic — but also commonplace. It’s practically the business model of all of Silicon Valley. It’s what “disruption” is built on. The playbook looks like this.
- Build a business by breaking a rule or convention that everyone else follows and winning over customers with a cheap or free offering.
- Rapidly scale to create a monopoly position (“blitzscaling“).
- Argue for regulation of competitors (this is called “regulatory capture“).
Oftewel: Do not disrupt the disruptor
Nog een mooie voorspelling van Bernoff:
Now that the American tech oligarchs are tight with the current presidential administration, expect a lot more of this. Just don’t get sucked in by the self-serving rhetoric.
Een voorbeeld hiervan is Mark Zuckerberg, kijk maar eens naar dit ‘interview’ met Joe Rohan.
Als je vanavond toch wakker ligt, probeer dan eens te bedenken welke voorbeelden er binnen het recruitment domein bestaan…