AI – Maverisk / Étoiles du Nord

Exit 2018 1163

Asad day for all you aficinados of this blog. After some five years and about 1163 posts (you’ll see…; own, mostly with own pics), this is the last of the (work)daily update. Yes, I’ve managed. But will turn to more serious, somewhat-more long-form content so will stop the drivel. I will not post daily, but when I do … And I’ll intersperse with some margin notes posts. Per 1/1 these will have no picture, the long for ones will – just check the link-post’lets and you’ll see. In line with the season: Enjoy less frequent but more professional, beautiful fireworks.
Or be safe with your own fireworks. Else, stand candidate for the Darwin Awards, which is also OK with me especially if you’ve not appreciated my blog; excepting the few I care for ;-|

Now then …
[Some room available. Live and die to be worth it, or take a hike; Arlington]

Some notes on notes on Chollet

After you read this, you’ll get the following:

[After two empty lines] ‘seed AI’ may not be necessary. Think of how the Classics built their arches: The support may be removed. Same here; some ‘upbringing’ by humans, even opening the possibility of ethics education / steering;
Proponents of this theory also regard intelligence as a kind of superpower, conferring its holders with almost supernatural capabilities to shape their environment / A good description of a human from the perspective of a chimpanzee. – correct. As such, slightly ad hominem and we know what that is about (here);
If the gears of your brain were the defining factor of your problem-solving ability, then those rare humans with IQs far outside the normal range of human intelligence would live lives far outside the scope of normal lives, would solve problems previously thought unsolvable, and would take over the world — just as some people fear smarter-than-human AI will do. – an interesting argument, as I had the idea of drafting a post about a new kind of ‘intelligence’, apart from the human/animal one.
Etc.

An interesting and profound read… Plus of course:
[“Intelligence”… Winter Wonderland London]

There we have it; botcracy

As we turn the leaf towards a new year, let’s not forget what values – in operation, operationalised – protect our Human Rights, in the form of de-mock-racy, and how they are ever so quickly being repelled by, e.g., AI and fake news but in particular, the deployment of bots as here.
Yes I know, that’s three layers of tools but still, the focus is on the first two but the latter plays almost the foulest role.
Yes I know, the ‘operationalised’ part may need elucidation on the side of ‘transparency’, ‘access and inclusion’ etc., but when you read after the link, you’ll understand that the issue is society-wide, not just FCC / net-neutrality.

Well, that was a quicky… hence:
[München, for zero (as in: 0.0) reason]

Trust ⊻ Verify

You get that. Since Verify → ¬Trust. When you verify, you engender the loss of trust. And since Trust is a two-way street (either both sides trust each other, or one will loose initial trust and both will end up in distrust), verification leads to distrust all around – linked to individualism and experience [we’re on the slope to less-than-formal-logic semantics here] this will result in fear all around. And Michael Porter’s two books, not to mention Ulrich Beck in his important one. So, if you’d still come across any type that hoots ‘Trust, but verify’, you know you’ve met him.

Since the above is so dense I’ll lighten up a bit with:
Part of the $10 million I spent on gambling, part on booze and part on women. The rest I spent foolishly. (George Raft)

Which is exactly the sort of response we need against the totalitarian bureaucracy (i.e., complete dehumanisation of totalitarian control – which approaches a pleonasm) that the world is sliding into. Where previously, humanity had escapes, like emigrating to lands far far away, but that option is no more. Hopefully, ASI will come in time to not be coopted by the One Superpower but then, two avenues remain: a. ASI itself is worse, and undoes humanity for its stubborn stupidity and malevolence (the latter two being a fact); b. ASI will bring the eternal Elyseum.
Why would it b. ..?
And if it doesn’t arrive in time, a. will prevail since the inner circle will shrink asymptotically which is unsustainable biologically.

Anyway, on this bleak note I’ll leave you with:

[Escape from the bureacrats; you-know-where]

The logic of automated decisions;
ransparency through audits ..?

Not bashing, nor FUDhyping…
Was triggered by various treads, e.g., The Book on the subject (or, het boek in Dutch), and scores of elucidation (yes. be happy finally there is some truly) from the legal perspective, on GDPR article 15.1h and article 22.

The latter two not being conclusive, however. They are about requirements of transparency on the logic underlying automated decisionmaking. But there is no clarity about how deep that should go. Will “Hey your data is processed by some AI system [literally, factually incorrect statement because it’s only Machine Learning at max, today; does that construe a false statement i.e. fraud ..? ed.] and even we the builders ourselves have no clue what goes on in there – that’s the whole point of using it besides being able to fire a great many inherently expensive humans and we don’t care the least about the biases and other grave errors of the system it works fine for us!” be acceptable? Hint: No. Will “Oh it’s so intricate that we, let alone you, have no clue when looking at the audit trails that the system generates” fly? Same hint.

Because here, we see a new area developing for IS auditors: Auditing ‘AI’ [quod non but read ‘ML’ and you’re good; ed.]. As IS auditors are (supposed to be, I happen to know a fair share of peers … etc.) the experts in gauging systems functioning qua .. reliability overall, too. Which goes way beyond mere C-I-A but still, has Always been part and parcel of IS auditors’ education, right ..? I will come back to you soon, with more definitive info on how IS auditors should go about this all.

Oh by the way yes I did already notice that the more the system in scope behaves, and is constructed to behave, intelligently like the average (sic! statistically you have zero reason to put yourself above that! oh wait you read my blog so you are definitely, way off the right end of the scale) human, the more the audit will have to be like we audit humans today. Uniting psychoanalysis and explicit rules on paper (in procedures, algorithms et al.), very dogue much fun.

Plus:
[Though a flat, and has iron, legally misidentified as flatiron …; NY – Pic tilted to fit in the pic frame of course]

Losing your trade’s virginity

I‘m referring more to dull trades, like auditing, than what your first thoughts were about…
It seems hard for some people to get their heads around the still persisting problem with AI introduction into regular trades, that when deploying AI to take over the rote grunt work at the lower rungs (which is, by definition (?), all that’s just behind your heels) and leaving the more intricate, ‘difficult’ and ‘intelligent’ work like decision making and risk weighing to seasoned, experienced professionals (to which you belong of course), that there will be no more seasoned, experienced professionals since the seasoning and experience is in years and decades of the rote grunt work that no longer exists for humans.
The ‘difficult’ decisions will all the more speedily be taken over by exponentially self-improving-on-the-intelligence-parts AI, as humans fail ever more quickly at those tasks. The excuse that the lower rungs failed in providing proper intel, will not work; higher-up humans would need to get a grip on the lower stuff, and to be able to determine the effectiveness of what goes around there… again you’ll need the extensive experience, maybe even more…

[Don’t get me started on how current ‘leadership’ (those that fell upwards by lack of weight, not the real leaders) already fails comprehensively at the intelligence part…]

Quite a vicious circle. And:
[Museum of what lies ahead for humanity, in total surveillance states, and AI futures; Riga]

Dumb judgement

If this (link in Dutch) is the state of the profession, then we’re all doomed. Luckily, the players in this sham [that’s putting it mildly, 007; ed.] will be deleted from history first. Sorry, not luckily; hopefully. Since the comparisons they make, and the judges’ explanations, are so utterly stupid that one can hardly see them function normally in regular society. Can’t sugarcoat this.

Those hat apply the law, aren’t above it I hope. Let appropriate parties get them, before they destroy communities and common sense.

Oh well. And:

[The circus is where such people were put on display, then the delusional got control; Zuid-As Amsterdam]

AI Blue-on-Blue

We keep on hearing these great things about how AI will help us in the battle against no-gooders qua information security. Like, in hunting for bugs in software (as asked for here, borne out in various much more recent cases or rather, news items hinting at pilot prototype vapourware) or hunting for fraudsters, possibly hiding in plain sight (superrrintelligent anomaly detection; unsure how false positives / false negatives are handled…).
Where on the Other side, great strides are also feared to be made. Deploying AI to improve (better fuzzify) attack vectors, and help with improvements in evasion and intelligence gathering in various other ways.

Pitted against each other …
When you know what Blue On Blue stands for (first of this), you will now see it coming, inevitably. What if autonomous (for speed of response!) retaliation kicks in …?

Never mind. I’ll like the fireworks show. Plus:

[Yeah, yeah, ships are safe in harbour but that’s not what they’re made for – I’ll just enjoy this view from a truly excellent restaurant; Marzamemi Sicily]

Nationalistic AI fuzzing

No his is not about fuzzing data. It is about accidentally giving away that you don’t understand a subject, and not the stats involved.
It is about this report. That was reacted on in various press – though not nearly enough and I don’t even have shares – by some boiler room country-by-country comparison, even without much of conclusional calls to action, for all…

Also, hardly anyone notices the gross error in it all. Which is the lack of proper definition of ‘AI’, or more expectedly, the widespread panicking brainfreezes of the interviewed.
Which, summa summarum for brevity, created such massive distortions that the figures are grey noise at best.

“Isn’t that harsh ..?” Nope. I did some asking around, for a different purpose, and when even 1% of organisations would do anything with AI yet, that 1% would be rounded up. Ppl were just too afraid to tell the interviewers / pollsters that they had (have, probably) no. single. clue., and babbled their way out of it.

So, what was this title again of the infamous Public Enemy hit ..?
And:

[A prettified prison is your ASI future; Zuid-As Ams]

Gödel around the White House

Any consistent formal system F within which a certain amount of elementary arithmetic can be carried out is incomplete; i.e., there are statements of the language of F which can neither be proved nor disproved in F.

How does that not relate to what we’ve been hearing a lot about in ‘Murican politics, lately ..? And, how does this not lead to absolute nihilism all around ..? The seriousness of purpose not reflecting the seriousness of the attack which is at most an amusing re-calibration of sanity, outside the attackers’ circles. But then again, how is that not like about-a-century-old politics, knowing where that lead to ..? If only the damage could be so-very-limited in comparison, to what humanity had back then.

But hey, maybe we will be burnt off the earth before anything that bad happens, right? Always look on the bright side of life [no link to the #1 of your playlist necessary]. And:

[It’s only a model”– also without link needed; ineffective as such but hey, it’s Châteauneuf itself; no “-“, near the church of Flip and Sjaak]