Are AI models learning to generalize?

We begin this episode with Paul Rosenzweig describing major advances in teaching AI models to perform text-to-speech conversions. Amazon flagged its new model as having “emerging” capabilities in handling what had been serious problems – things like speaking with emotion or transmitting foreign phrases. The key is the size of the training set, but Amazon was able to pinpoint the point where more data led to unexpected skills. This leads Paul and me to hypothesize that training AI models to perform certain tasks ultimately leads to the model learning to “generalize” its capabilities. If so, the more we train the AI ​​on a variety of tasks – chat, text-to-speech, text-to-video, and the like – the better the AI ​​will be at learning new tasks, as generalization becomes part of its core competency. We are advocates championing the frontiers of technology, so take this with a grain of salt.

Cristin Flynn Goodwin and Paul Stephan join Paul Rosenzweig to provide an update on Volt Typhoon, the Chinese APT that is peppering Western networks with the equivalent of logical landmines. It’s actually not so much an update on the Volt Typhoon, which appears to be aggressively pursuing its strategy, but rather the hyperventilating Western reaction to the Volt Typhoon. There is no doubt that China is playing with fire, and that the United States and other cyber powers should freely seed similar weapons into Chinese networks. Unfortunately, despite the panting, the public measures adopted by the West do not seem capable of defeating or deterring the Chinese strategy.

The group is unimpressed by the New York Times’ claim that China is waging a dangerous election influence campaign on U.S. social media platforms. The Russians do it better, says Paul Stephan, and they don’t do it well either, in my opinion.

Paul Rosenzweig examines the House China Committee report alleging a link between US venture capital firms and Chinese human rights abuses. We agree that Silicon Valley VCs have paid too little attention to how their investments could undermine the system on which their billions rest, a situation that likely won’t last long. Meanwhile, Paul Stephan and Cristin update us on US efforts to disrupt Chinese and Russian hacking operations.

We will eagerly await the resolution of the European battle over Facebook’s subscription fee and websites’ implementation of “Pay or Consent” privacy terms. I predict that the hypocrisy of the Eurocrats will be tested by the effort to reconcile the sentences for elite European media sites, which have already embraced the “Pay or Consent” principle, with an almost obvious sentence against Facebook. Paul Rosenzweig is confident that European hypocrisy is up to the task.

Cristin and I explore the White House’s latest enthusiasm for software security responsibility. Paul Stephan explains the difficulties related to the UN treaty on cybercrime, which is and should be blocked in Turtle Bay for the next ten years or more.

Cristin also covers a detailed new Google TAG report on commercial spyware.

And in quick shots,

Download 492nd episode (mp3)

You can subscribe to The Cyberlaw Podcast using iTunes, Google Play, Spotify, Pocket Casts, or our RSS feed. As always, The Cyberlaw Podcast is open to feedback. Be sure to interact with @stewartbaker on Twitter. Send questions, comments and suggestions on topics or interviewees to CyberlawPodcast@gmail.com. Remember: If the suggested guest appears on the show, we’ll send you a highly coveted Cyberlaw Podcast mug!

The opinions expressed in this podcast are those of the speakers and do not reflect the opinions of their institutions, clients, friends, families or pets.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *