AI: Scaling AI Trust is Job #1. RTZ #382

...Microsoft et al learning by trial and error

The Bigger Picture, Sunday June 8, 2024

As I’ve discussed at length at this early stage of the AI Tech Wave, companies racing to scale AI products and services have a unique challenge vs previous tech waves. Addressing users’ ubiquitous lack of Trust that AI providers will protect their content, data, and privacy. Scaling user Trust in AI is as important as scaling AI itself, a point I outlined this week discussing Apple and its upcoming AI announcements on Monday. This issue has resurfaced again this week in the context of Microsoft. It’s the “Bigger Picture” I’d like to focus on this Sunday. Let me explain.

But first, the Microsoft context, in terms of the backlash over its AI Copilot PC Recall feature. Microsoft decided to turn it off by default. It’ll now be an ‘Opt-in’ feature when the new Microsoft AI PCs roll out in the coming weeks.

As CNBC explains in “Microsoft says AI feature that captures screenshots on new PCs will be off by default after backlash”:

“Microsoft said an artificial intelligence feature on new PCs that captures screenshots and enables searching of user activity will be off by default after security researchers determined that attackers could access the underlying data.”

“The Recall feature was one of the main capabilities Microsoft showed during a press briefing last month for forthcoming Copilot+ PCs with AI computing power onboard.”

“If you don’t proactively choose to turn it on, it will be off by default,” Pavan Davuluri, Microsoft’s head of Windows and Surface devices, wrote in a blog post Friday.”

“Microsoft has been trying to balance competing interests of late as it moves to incorporate new generative AI tools into its products and to keep up with the competition. While the market is evolving rapidly, user privacy and security are under a microscope.”

This was a rapid turn of events since Microsoft announced its signature Recall ‘ feature for tens of millions plus of new Windows laptops, and tablets to come: Recall. The promise was instant and magical memory of everything the user did on the machine, with a promise of total, local security of user data.

And all with ground up re-building of Windows and ‘local’ AI models (aka small language models or SLMs). Dubbed AI ‘Copilot + PC’, these first wave of machines are to be equipped with the state of the art Qualcomm Snapdragon Elite X low powered chips, all powered with AI-optimized ‘NPUs’ (neural processing units), to crunch crazy amount of AI calculations locally.

I outlined this in a piece on how Apple was likely best positioned on this ‘Trust’ dimension vs other big tech peers. And that’s because of the company’s unique focus on user ‘Privacy’ for years, driven with a laser focus by CEO Tim Cook on down.

Ironically, for AI to be truly useful it’s going to require our devices to constantly watch and record everything we do. Then process all that data with ever scaling AI hardware and software, to deliver uniquely useful, probabilstic ‘Reasoning’ results vs traditional deterministic computing systems we’ve had for decades.

Be they laptops, tablets, phones, wearables, smart ‘Voice’ devices like Apple Siri, Amazon Alexa, Google Assistants and so many more. In the tens and soon hundreds of millions.

The rub is that billions of mainstream users are going to have to Trust the companies providing these services. And given the nature of the business models in tech to date, primarily driven by advertising, and the notion that the ‘user is the Product’ for advertisers, that Trust is going to be hard to come by initially.

All the big tech companies have a long way to go to earn user Trust as AI truly scales to its potential possibilities. A singular challenge in this AI Tech Wave, is going to be to earn that Trust at scale. And that is potentially going to take recalling a lot of prior practices at scale. Many mainstream users are not going to just take ‘Trust Me’ as their word.

That is the new AI “Bigger Picture” ahead of us. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)