Op-Ed: Mixed messages from and about Microsoft are fuelling AI doubt on a different bandwidth

Of all the big players in AI, Microsoft is the one with the direct hardwired connection to the mass global market for hardware and software. On the subject of AI, Microsoft is pushing its Copilot and related packages hard, while also generating a lot of static about AI risks and realities.

The messages vary, but the room for doubt is expanding exponentially. These are current headlines over the last week:

AI could create a ‘silicon species’ that rivals humans, Microsoft chief warns. Sounds helpful. This includes the usual “impossible to control” mantra.It’s also far from impressive, given that these are the people responsible for controlling it, and the only people who can control it.

Microsoft AI Chief Sends Stark Warning on AI Safety. Same basic message, slightly different angles.

Microsoft exec called AI scraping the “largest theft of labor in human history.” Another resounding ambiguity on the subject on the legitimacy of basic AI processes. AI is obligated to scrape, not think. It can’t do anything else. The question is whether it or the LLMs, or anyone else, learns anything. Anyone who knows their stuff can see where it came from and judge for themselves the value of AI-generated outputs.

The somewhat less flattering Microsoft Patches CVSS 10.0 Azure AI Foundry Flaw Enabling Unauthorized Privilege Escalation adds a practical dimension showing real-time major issues for users.

New Microsoft AI Code of Conduct Emphasizes Human Control. This relates to the Microsoft Code of Control, a foundational policy document that simply emphasizes the general gut reaction to AI in its present form.

Does any of this strike you as any grounds whatsoever for confidence in AI?

Doubts multiply into sub-doubts and tangential doubts. It’s an odd look. It may be honest or not, but it’s also not showing any Exit signs. These messages translate into a mess.

Microsoft background

Microsoft has a history of consumer blind spots. Typically, replacing a perfectly good OS that everyone was OK with is a problem. Glitches and weird update issues add little to this sour mix of issues. Ending support for older systems when that support was already in place and automated, etc., etc., is simply annoying.

Given this history, introducing a whole new class of tech into the consumer market, whether anyone likes it or not, could also be a hard sell.

There’s a point or two to be made here.

The consumer end is core business for Microsoft, with or without AI, and this market is far more valuable than any short shelf-life gimmicks.  

People simply do not need AI, let alone advanced AI, to do their own current business at the consumer level.

The Internet of Things doesn’t need it, either. Does your fridge want to be a stunt double for NASA? Maybe not?

You can easily argue that doing the same essential things with unknown issues and risks isn’t a selling point.

Consumers definitely don’t need AI third parties they can’t trust and easily manage.

People do not want to think about having to pay unspecified and unspecifiable costs for unknowns anyway.

It ain’t broke, so don’t fix it

It’s not just Microsoft. The tech sector has an unwholesome reputation for being dazzled by its own brilliance. It has an equally unentrancing reputation for delivering god-knows-what shoddy tech that barely, if ever, survives contact with the real market.

The unhesitating acceptance of AI, followed by years of debacles and vacuous commentary about return on investment, doesn’t add credibility to this picture of bliss. AI still has to prove real value, not just saturate the media with its brochures.  

Superimposing AI on the current business and consumer tech levels is really pretty absurd. Anyone with so much as a sniff at the most minor business qualifications could prove in seconds that “productivity” and “getting rid of built-in inefficiencies” are really the same things. You need organization much more than you need tech. Hardline C-level people with budgets need convincing, not schmoozing.

The digital revolution simply bulldozed the paper inefficiencies. That won’t happen again. AI is not the same thing. It’s inherently capable of a much higher value tier of processing capability, and it’s being used for trivia.

At business grunt level, using vast amounts of capital commitment for AI compute to do the world’s most menial tasks is idiotic economics and utterly unnecessary. Taks value is a very appropriate metric for evaluating consumer-level AI. Current outlay is already murderous and can’t be propped up by the consumer market.

Direction is what’s needed

The reality of mass market AI adoption is likely to be stormy. The fan will be hit and instantly overloaded. Microsoft will be the first lucky recipient of massive market feedback for and against consumer-level AI, on a scale that obliterates current feedback.  

This will turn into a historic saga of market positioning and repositioning. Lack of clear direction and idiot-proof AI at the consumer level could be fatal. Even the big security issues could be lost in sheer systemic and network dysfunction if AI fails on this market level.

The one thing that’s gone right about these mixed messages is the fully justified skepticism. Everyone gets that.

Something called Windows AI Home Edition may be the opportunity to prove that Microsoft is still in touch, acting in good faith, and the same company that fired up digital tech worldwide.