To EU and E Warren: is a sledgehammer the way to fix google etc?

Finally people start to realize that massive changes in the world internet have just begun, and that massive problems could emerge if we don't develop a better system of cooperation. The EU has promised to develop a global regulation system to control AI, and Warren says she will just break up Google, Amazon and Facebook.

There really are many lethal problems moving fast, but do any of these folks have any idea how to fix them?

I keep being haunted by an image on an unshaven auto mechanic wielding a huge hammer, saying: "I ain't no engineer. I don't know how cars work. but I sure can fix YOURS but good.."

======

In parallel with that, there are a few discussions starting up about how one could do it right.
Here is one example:

A really crucial issue here is metrics, and truth in metrics.

My wife has told me more about "B" corporations, which do not HAVE to be organized around maximum

profit. I hope SEC allows that, not just some states. From what I have seen of oil company shareholder meetings in the UK (well publicized for good reason in FT), THEY have big and powerful shareholders who want a strong signal that a weighted sum of profit (however measured) AND climate change benefit is what they want, at a minimum, starting ASAP, because a lot of people might suffer if we don't get going, in the opinion of those (well informed) shareholders.

THIS IS NOT JUST a cause in itself, but a great TESTBED example of structural changes needed to really accomplish stuff like that.

For IT folks:

What would be needed to allow creation of a new corporate "spreadsheet" (or"blockchain" or certified ledger) with entries which predict LONG-TERM benefits to reducing risk for climate change?

Short term measures just won't do the job (e,g, allow sustainablepolicies to results), because myopic measures like the number of solar panels on roofs tend to force BAD, counterproductive decisions. (For electricity in Europe, better than solar panels in houses in the north would be big well-designed solar farms in southern Europe hooked up by new transmission, giving Germany the HUGE side benefit of redirecting money for loans or governments to money which buys them electricity and creates real jobs in the South.) 

=======

SINCE the world IS going to change very massively in the next decade or two, due to new IT far beyond the wimpy zero generation "AI" which you see in the press these days, we need to think more clearly and concretely about what the possibilities MIGHT be like,

grounded in the more advanced new technology. 

I will soon recommend that anyone interested in these transitions read Schoeder's new sci fi,Stealing Worlds. I am only 100 pages in, 

and may or may not end up agreeing with the final conclusions (even he misses some important stuff), it includes many possibilities which we need to account for. I was amused by discussion of a new platform (used by only a few companies) where an AI runs an entire company instead of the CEO. (After all, of shareholders do not NEED a CEO...).

No, that would not be a face recognition program. (Some versions of Watson IOT sound exactly that dumb. My slides at www.werbos.com/IT_big_picture.pdf depict a few options relevant to corporations and larger systems, which some folks tried to take over big parts of the US government, folks who still try.) It would be a decision-making system. Alpha Go is a very simple decision system, but MUCH more powerful are in the works, some in use.

But good decisions by human, by AI, or by some kind of conglomerate (designed how?), require FORESIGHT. 

In fact, the time-series kinds of networks which can offer that (with probabilities and scenarios etc)are important to HUMANS as well, since they are crucial to accurate "state identification," and even to humans learning what was FOUND in massive data trawling exercises. 

If people's data disappears into a black hole, and is not input to networks trained to output TRUTH... there are many sinkholes ahead of us in the unregulated IT world. Is truth essential both to automated corporate spreadsheets and to preserving the full rights of consumers and workers? And how do we enforce truth in those neural networks or hybrid neural/human networks?

Of course, the management of short-term performance metrics (not the same as long term bottom line larger scale goals) is also a challenge requiring lots of training and truth.

Many old style AI folks want to go back to digital expert systems using Boolean words for the next big step. That won't work, but DIALOGUE with humans is a key element as well. It is an interesting question what LEVEL of AGI should be used/allowed for now in such functions.

Best of luck,

    Paul 

============

My old alumni association cites an editorial on the problem of truth in news:

https://www.wsj.com/articles/impartiality-is-the-source-of-a-newspapers-credibility-11568109602

This is just one of the many serious manifestations of the issue of designing truthful networks.

Of course, the folks who deduce things about YOU from YOUR data is also serious.