Friday, December 31, 2021

Grand Challenge

I just became aware of a talk given this year by Alan Kay where he looks at our evolution (humanity) and considers how thinking became more capable. Or did it? Well, yes, even though it's not apparent in many ways. Too, he proposes that machine learning being put first as we have seen with AI the past decade or so is like the tail wagging the dog. 

He suggests that we need a new platform and gives four basic notions all of which I agree with. Except, I would take it further, KISS notwithstanding. So, all of this is debatable. The platform, to be discussed as we go along, is the Grand Challenge. It runs down the same alleys as does truth engineering.

For now, here is a list of Alan's required resources: Folk Explanations, ... and Other Minds, Bertram E. Malle; the work of Doug Lenat; Causality, Judea Pearl; UI work, Bret Victor.  

FutureLaw presentation
Stanford Law School 

Platform? Lots to talk about there, as with the other themes involved with this important topic. 

Remarks: Modified: 12/31/2021

12/31/2021 --

Saturday, December 25, 2021


Foreword: I could say, Christmas present, as I just became aware of this a couple of hours ago. Oh yes, I mention reading old papers. The stack is there. I do get through things regularly. But, I also get more reading. Like, the end of the year? Nothing better than the double issue of  The Economist. Never miss that one. 


So, today, I was reading some older WSJ issues. I buy them and throw them on a stack for later reading several times a week. I try to get the weekend edition. The 15 Dec 2021 issue, there is an article that mentioned a book, by Graham Allison of Harvard, which looks at technology and war. The theme is whether the U.S. is losing to China.

Okay, they like to talk 5G, but it's still a work in process. I just got my first smartphone and have not spent much time on 5G. Would still be happy with 3G, but they took it away. So, 4G? It's been great. BTW, this is no luddite writing. I spent my career in advanced computing technology. Rather, I see things expanding like mad without rhyme or reason. 5G has a way to go; now, they're talking 6G (of course, mere whisperings). 

Let's talk one issue. So far, I have had two posts using a contraction: AIn't and AIn't, again. Bascially, after spending close to two years looking at AI, again, in an in-depth mode with a technical focus, I can see where the hype comes from. People talk. Some listen. Not many dive into the details. So, hype rises. 

It's human nature. Even techies do this. No one can spend the time to prove everything from first principles at all times and places. Just AIn't possible. So, we believe experts. And, we trust the government. This post is about that. 

After seeing the 15 Dec 2021 op-ed, I was browsing 8 Dec 2021. Same guy. The theme? Here are two bullets summarizing the subject mater. 
  • The National Security Commission on Artificial Intelligence - Wow. I was being technical. Thank God for that and not paying attention, except to watch the marketing hype and read on the angst. Big names. Funny, how this psychology is going to unwind itself. Well, gets your attention? That's 'National' in terms of the U.S. and included everyone (of course, DOD). But, they're not around any more. Since, there was a report. 
  • The Report (archived) - I don't know how long the [dot] gov stuff will be around. So, the archive is legit. Notice, the Final Report (was released on 1 Oct 2021). There is a PDF and other resources. 
  • Table of Contents - Thorough. As mentioned, lots of participants. So, we will look more closely at all of this. It's basically full of recommendations and suggested actions. 
But, on a quick read, my thought was 'AI is' by government decree. Well, we see lots of that. Or have. As mentioned, the psychology of this will be interesting to dive into. But, it's not set, yet. 

In the meantime, it's not unlawful to note that AIn't. I will switch to the 'Tis side when evidence comes about. There is none (prediction) in the report. Anything bordering on that would be rhetoric that can be challenged with a black board test. As in? Yes, let's step through the mathematics. For real. No quick and dirty. No, we have a couple of centuries to cover. 

Trouble is, I may not live long enough. I sure can talk truth engineering with regard to this. 

Remarks: Modified: 12/25/2021

12/25/2021 --

Friday, December 3, 2021

AIn't, again

 A month ago, we did a post on the subject of AIn't which is depicted in the graphic. Today, we offer a new view with coming updates. 

Some mention that there is a lot of hype, seemingly driven by marketing. For what reason, one might ask. Too, others have noted that the NN part of this, while creative, might be better done under some purview of a scientific framework. 

We usually do not live where experiments are done on the populace as a whole, except that has been the modus operandi of economists from its start. 

Big pharma needs to prove that which they will be pushing, at least, they have since the days of the snake-oil salesmen were put to a close. 

Remarks: Modified: 12/25/2021

12/25/2021 -- NSCAI and its report.