This is the script of my national network radio tech report last Monday on the topic of Meta's monitoring of employees' computer activities, using that data to train AI -- and discussion of the broader implications of such projects. As always there may have been minor wording variations from this script as I presented this report live on air.

- - -

So yeah while this discussion is about Meta the same trend is happening at other Big Tech firms where employees are increasingly being subjected to mass layoffs or feel that they are being mistreated by management, and the impact of this could be quite dangerous for everyone who depends on services from these firms as the firms try to use AI as an excuse for all manner of bad corporate behaviors.

But regarding Meta, it's useful to recall that Meta's CEO Mark Zuckerberg has never been the poster boy for particularly admirable corporate leadership. Of course before Meta was Meta it was Facebook, and before that it was The Facebook and before that it was Facemash. This takes us all the way back to when Mark started this entire sordid set of events when, while a student at Harvard, he built a platform to help guys avoid dating what he and his Harvard bros considered to be unattractive women, sometimes using terms like "farm animals" for their comparisons. Apparently he threw this project together after being dumped by his then girlfriend.

This was all back around 2003. Facebook didn't change its name to Meta until around five years ago or so, when Mark became convinced that everyone would want to wear virtual reality headsets, working and playing in his virtual reality "metaverse" all day, and threw reportedly tens of billions of dollars at that project, some reports say 80 billion or so. And while there are continuing to be amazing applications for VR and Meta VR headsets, the idea that this was going to be a dominant way for people to routinely interact was, what's that technical term? -- oh yeah, crazy.

Of course over the years Meta/Facebook has been embroiled in a wide variety of issues related to privacy and a long list of other controversies. So now Mark has changed his mind again, and has decided that he should throw most of Meta's resources at -- you guessed it -- AI. And the way he did this was for example, reportedly by reassigning product managers and skilled engineers to be essentially AI trainers, understandably considered to be work far below their skill levels, and laying off other employees. Then Meta came up with the idea that is somewhat akin to driving a stake through the heart of employee morale, already apparently at historically low levels for the firm. And while I don't know offhand who specifically originated this train wreck of a concept, it seems like something Mark would have quickly and enthusiastically endorsed.

So this is called MCI -- Model Capability Initiative. This was launched just a couple of months ago, and collected employee keystrokes, mouse movements, screen shots, etc. to be used for AI training. Initially there was no opt-out. Then they apparently introduced a limited half hour opt-out. As you MIGHT imagine this didn't go over well with most employees. Even when using a company provided computer as in this situation, employees routinely deal with personally sensitive subjects including financial and medical information and much more. And you can't blame employees for being decidedly unenthusiastic about being forced to train AI systems that they quite reasonably suspected Meta would like to use to lay them off and replace them.

Well, this was all bad enough, but then very recently the fan really got splattered when it was discovered that the data being collected this way had not been properly secured internally, and was reportedly accessible to pretty much anyone at the firm. Whammo. So about a week ago the MCI program was suspended for some indefinite period -- maybe forever, we don't know -- while "investigations" take place into how this happened. This whole awful affair seems pretty much on-brand for Zuckerberg, but again it would be a mistake to think that this kind of situation could only occur at Meta.

Because across the technology sector we see firms increasingly treating the employees who actually built these firms as expendable, sometimes to be replaced with half-baked, dangerous misinformation-laden AI, that drives customers and employees alike utterly nuts.

But it does seem like Large Language Model AI is close to "jumping the shark" as the saying goes, or perhaps already has, and is starting to face some serious reckonings. That recent German court decision I discussed recently, holding Google responsible for the content of their AI Overviews -- that spout 10s of millions of wrong answers an hour -- could be just the beginning.

If courts begin widely holding Big Tech AI firms responsible for their AI-generated content -- search answer overviews, chatbots, and so on, this very reasonable, common-sense approach could trigger massive changes in the way these firms operate and might signal an end to the abandon with which AI-created misinformation, sometimes dangerous and harmful, is so widely spewed.

How these firms treat their own employees could be viewed as somewhat akin to a canary in a coal mine regarding how these firms will treat all of us. And judging from the mess at Meta in this respect, we probably shouldn't expect a "customer is always right" approach from Big Tech firms, that's for sure.

- - -

L

#meta #google #ai #metaverse