Saturday Links: Google Enterprise AI, Training Data, and Robot Rescue

Here are this week's links:

• The "Everything AI" Google Next Keynote. Google went all in on AI in its showpiece keynote this week. Its core pitch was that it has the deepest infrastructure base to deliver AI both to the consumer and to the enterprise. There were some impressive product notes. I'm still wondering how Android and search will ultimately be affected by AI (and the options it opens for competitors), but on the Enterprise side, Google is pulling out all the stops. From the demos, its pitch is essentially that it can do everything: build models on your data, create customer service bots, infuse product design, and much more. Swathes of startups are taking on each of these verticals; Google will be a tough competitor.
• The New York Times delves into AI Audio. The NYT aims to add generated voice audio for all its text content over time. This modality has been available for some blogs and other news outlets for a while (sometimes with human readers) and it can be a great new modality to consume content: it's easier to listen to an article than read it while on a run. In the bigger picture it matters too. The NYT is, I believe, the largest online subscription news service. Audio now, perhaps multi-media video later. Writing the text opens the gate to becoming a media channel across all modalities.
• Robot rescue (Adagy - Y-Combinator W24 Batch Company). This company caught my eye among Y-Combinator's latest batch of companies. They aim to provide a support service that can remotely log into robots (mostly industrial / farm robots) and restart them / fix them. It's funny to think we'll soon be in a world where you need an AA/RAC+IT support service for your autonomous systems. Possibly somewhat ahead of its time but prescient. (A good colleague pointed out that many industrial machinery manufacturers will probably fight third-party maintenance tooth and nail.)
• Meta ‘discussed buying publisher Simon & Schuster to train AI’. As the search for training data heated up last year, Meta apparently considered buying publisher Simon and Shuster as one possible avenue to get more proprietary content. It makes you wonder whether there is a cultural tipping point in the future where human authors produce work where the larger chunk of economic value is actually in training AI to generate new works (or just do their jobs) rather than entertaining humans directly. Also... what happens when AI has all the training data it needs. I doubt this will actually happen in a real sense (humans+AI will be the winning combination for a long time), but the prospect of Meta owning a publisher (and that making economic sense) is a strange one to get your head around.
• Suno. For the past few weeks, Suno.ai (now Suno.com) has been popping up in emails and Twitter everywhere as the new generative sensation for full music tracks. If you haven't tried it yet, it's worth doing so - the results are surprisingly good. Two random prompts from me: "A melodic psy trance track about open water swimming" and "A death metal banger track about motorcycle racing in death valley".

Lastly, if you are an ATT customer in the US, then in the past ten days, your personal details, including your social security number, were probably leaked due to a hack. Pay attention to the ATT emails in your inbox about securing your account. There's not a lot you can do, but they are offering Experien Identity Works access for free, which might help a little.