The Conversation here on the place of understanding in AI – effectively defining AI as not involving understanding. That is Intelligence Without Understanding. Not intelligence in my book, mere competence in Dennett terms, but here Sandis and Harper are suggesting there would be too many snags if “AI” did have understanding, that it would be a bad thing.

My position is that kind of AI is just algorithmic automation, not actual intelligence. Clever and useful but not intelligent. This is simply the AI-hype making a public song and dance about the kind of industrial systems automation and control that’s been around for decades, a century, and is simply becoming a more apparent (more hidden) part of applications on the public web. In my book that’s not intelligence, and part of the reason I say that AI will not be intelligent until it evolves to be actual intelligence. Proper-AI  will only happen when it has evolved through life itself. I happen to believe that is possible, that machines sufficiently complex over long enough times in stable enough environments will evolve “artificial” sentient life and ultimately actual artificial intelligence. But that is a very long way off.

This is just a definitional problem – people with interest in the public hype of AI finding it useful to redefine intelligence broadly as anything unintelligent – something that merely quacks like a duck. Those of us interested in human intelligence hold intelligence to much higher standards – genuine scientific and philosophical standards.

On the other hand, the trouble is that so many so-called science types – in order to stave off any hint of the thin end of a supernatural wedge – seem keen to redefine human consciousness and will as mere illusions, that even human intelligence is automatic. This is a political dispute – a war – not involving any actual science. This warlike political dispute with winners and losers and cost-benefits is what sells media clicks and eyeballs, to the disbenefit – polarisation – of science. (*)

But here, Anil Seth speaking at TED earlier this year:

Interesting that Nigel posted this after I had responded with a different spin to his own conclusion on human intelligence being a bleak vestige of possibility being pushed-back by neuroscientific advances.

That alternative spin is in the footnote here Irrational Science Portends Inhuman Transhumanism.

Anyway, watching Seth talk, we find that he is a scientist who is concerned about the intelligent and willful realities of human consciousness. I tweeted a few reactions:

Telling that, as a scientist, he makes an “even some philosophers” gag, which is missed by the science-worshipping audience. But there are plenty of cynics in his Twitter timeline – here an example:

That “lol” says a lot about how unscientific – how dismissive – the cock-sure scientistic types are when philosophers and sophisticated scientists suggest there might be things beyond their Science-101 comfort zone.

I like Seth’s attempt at a scientific ontology of consciousness – one that genuinely recognises our subjective position. The sexy graphics and use of the term “hallucination” grate a little, but are the necessary click-bait I guess. There is an honesty – that “sorta” operator – in how it’s actually described. Seth limits himself to what intelligent consciousness is, and importantly its dependence on life and embodiment, and doesn’t delve into the evolutionary philosophy of how it possibly came to be, but there is enormous alignment here with Dennett.

[Separate post to follow – gutting a summary of that ontology from Seth’s presentation, and relating it to Dennett’s evolutionary process view.]

=====

(*) Interesting debate (in two parts) on BBC R4 Today this morning – Steve Jones, David Willetts and Richard Dawkins on science in the media and in public political life. Jones eventually acknowledging that most media coverage of science is about its technological and economic exploitation and mis-representing political controversy as scientific controversy. The politician talking the most sense – most recognising that not only is politics not science it’s not even about science and certainly not restricted to questions of science and evidence. Science may – rightly – wish to exclude values and subjectivity from its processes, but it cannot dismiss them from public interpretation, debate and controversy over content and exploitation of science.

[Post Note: Lots of Anil Seth resources linked on his blog page. Wider than the sky!]

I find myself bookmarking or retweeting almost everything from EES because their agenda seems to coincide with so much of the thinking I’ve been doing here on Psybertron for the past 15 years and for a decade or more before that.

Falling into that recurring trap of having so many things bookmarked – deserving of deeper reading and consideration – but which inevitably postpones the necessary reading and writing.

This is no different. I don’t know much about “who” EES are, all I know being from the content of their site. When it comes to “isms” I’m one for “good fences make good neighbours” rather than hard dividing lines, so I’m not overly concerned with exactly which thinking is merely the “modern synthesis” and which is properly the “extended synthesis” when it comes to actual understanding of evolutionary processes and possibilities. The distinction to me feels a little like my own self-labelling as PoPoMo to distance from prejudice about PoMo, whilst nevertheless seeing latest credible thinking as building constructively – synthesising – on what has gone before.

More reading to be done!

Just a bookmarking post.

Hat-tip to @anitaleirfall for posting this link to Daniel Kahnemann’s “retraction” of underpowered statistical significance data in his Thinking Fast and Slow based on his earlier work with Amos Tversky.

Noted previously that super-statistician Taleb had profound effect on Kahnemann.

Kahnemann’s work is behind “Nudge“.
(He and Tversky much referenced in Thaler and Sunstein’s eponymous book?)

Taleb had been involved with Nudge with UK as well as US governments (The David Cameron story?). AND Taleb had indicated the problem with Nudge when, as so often, the wrong or false facts were used as the basis of Nudge, or unintended consequences resulted.

Does the technical error – which changes the significance of the facts – actually change the implied / accepted reality, prove it wrong or simply leave it unproven?

Many meta-meta-levels in this …. me internalising Kahnemann’s error – doing so correctly – making the right Kahnemann / Taleb / Nudge inference(s) – and the question of how significant these might be … as Nudges. (There’s a lot more in Thinking Fast and Slow – not all dependent on the above error.)

The real dilemma here is small facts
having much greater significance than might appear,
despite seeming insignificant and easy to internalise implicitly
– non-contentious, easy fit with prejudice –
yet being significantly wrong!

Story of our lives?

(Of course the opposite case exists too – as Kahnemann himself admits – failure to internalise a significant correct fact because it’s insignificant appearance.)

References to Yuval Noah Harari’s “Homo Deus – A Brief History of Tomorrow” are coming thick and fast:

Andy Martin in The Independent “Transhumanism: The final chapter in humanity’s perpetual quest to be kitted out in comforting accessories.” (Previous brief reference solely on Information aspect).

Philip Kitcher in the LA Review of Books “Future Frankensteins: The Ethics of Genetic Intervention” (ht @KenanMalik) which is combined with a comparative review of “A Crack in Creation – Gene Editing and the Unthinkable Power to Control Evolution” By Jennifer A Doudna, Samuel H Sternberg.

Rory Fenton in the New Humanist – “Will progress kill humanism? – the idea that scientific knowledge might one day undermine democratic values.”

I still need to read Homo Deus and digest all three reviews, but initial thoughts are as follows:

In reverse order starting with Fenton in the NH, there is indeed a risk, one that exercises me daily, that science is indeed in danger of undermining our human values and democratic freedoms. But that is a narrow ill-conceived, an arrogant populist kind of science; a too reductionist, too objectively deterministic conception of science. It’s a dogmatic ideology I tend to refer to as scientism.

Fenton summarises one aspect of Harari’s position with the following:

“For humanists, free will is absolute, our sole driver. But advances in both neuroscience and computers are undermining this view. Harari cites experiments that seem to show a decision is made before the person is actually conscious of that decision, and fascinating experiments with people who have had their left- and right-brain hemisphere disconnected, who will then justify the same decision with different logic, depending on which hemisphere is being stimulated. Advances in computers that make machines intelligent, if not conscious, leave more scientists convinced that a similar computer-like process must govern the mind. Harari paints a picture of the brain making decisions automatically, which the conscious mind then justifies and takes credit for.”

At this point I can’t be sure whether this says as much about Fenton as it does Harari, but that paragraph captures both the dogma and the incoherence of the scientistic position. In doing so, we would be right to fear a disaster if machine automation were permitted to embody such a flawed scientific conception of reality. Those flaws become out of sight, out of mind, ignorantly accepted and ever more remote from human correction with the increasing pace and scale of automation of machine-learning, processing and control, and their embodiment in invisible layers of algorithms within the web on which we depend.  True rationality, true free-thought-based Humanism, needs to get a grip on the the reality of flaws in our scientific model before it accepts their mechanisation.

An irony often mentioned here is the implicit importance of democratic freedoms to “free-thought” humanism – maybe not “absolute, sole driver” as suggested above but pretty fundamental – yet falling hook line and sinker for the determinism of science banishing our conscious free-will from the functional picture. It has become a cliche to cite the Libet (and other) experiments supporting “a picture of the brain making decisions automatically, which the conscious mind then justifies and takes credit for” in support of that ludicrous position. (Much written about here, see also post-note.) Neuroscience and information science are undermining human freedom only because they are reinforcing the flawed – objectively deterministic, reductively scientistic – dogma rather than actually advancing. That dogma is in danger of becoming a bar to the self-correcting evolution of science and rationality itself. Let’s not automate it before we fix it by freeing the dogma.

Malik tweeting a reference to Kitcher’s review refers to:

“The superficiality of Yuval Noah Harari’s post-humanism.”

Amen to that. In fact it’s a summary of a quote from Kitcher “The gods glorified in the post-humanism of Homo Deus are capricious, superficial, and cruel” Malik’s thinking is generally nuanced and high quality, so it would seem to bear out my position on the rather pale imitation of scientific rationality being presented by the science offered.

Fenton is non-committal on his own position, simply presenting Harari’s warning:

As [flawed] science and technology undermine concepts of free will and a true inner “self”, Harari foresees a threat to the prospect of a world in which we value the uniqueness of each person and trust them to make their own lives. This is not something he necessarily welcomes; rather, his book serves as a warning about where we might be headed.

The reason we may be headed where we humanists should fear to tread is because the science we subscribe to is flawed in a dangerously dogmatic way. The content of science may always be contingent and self-correcting, but this is a more systemic problem we need to address directly. What makes science rational?

In concluding, Kitcher says:

“Readers of Homo Deus wait in vain … for a clear recognition of what has been achieved and a sensitive reflection on how it might valuably be employed. Harari’s stampede to the post-humanist future is unchecked by ethical ruminations.”

“Humanity surely needs more grown-ups.”

That last phrase nicely captures my problem with the populism of Science-101. Good science needs to grow up and respect a wiser view of rationality.

[To be continued, more reading to be done.]

I suspect it may turn out as Martin commented, that “Harari’s Homo Deus is endlessly fascinating.” Humanists who worship the rationality of public scientists need to be just as vocal supporting public interest in the humanities.

=====

[Post Note: Nigel Warburton writing on more android than generic AI, also invokes Libet:

Libet himself left some room for control.

He suggested we can think of ourselves as having “free won’t” rather than free will …

Whether or not he was right, the thrust of much recent neuroscience is that far more of what we fundamentally are occurs beyond the control of our conscious mind …. a bleak picture of what it is to be human, but it may be accurate. Perhaps we are closer to [the robot] in some respects than we might like to think.

“Closer; Far more; In some respects”; all carefully qualified, but odd again to find the philosopher so open to the bleak conclusion, even though he brings in many other sources that I use here. I too invoke “free-won’t” as the better model. Far from being bleak, the recognition that our conscious will is very small compared to our many layers of pre-conscious action is encouraging evidence that our consciousness really is an emergent, evolved capability of an intelligent, sentient mind. Furthermore it is evidence that it has evolved to be efficient. The kind of consciousness worth having. It’s the greedy reductionists that see “small” as some minor skirmish standing in the way of determinism’s total victory over humanity.]

Despite writing much about this before (in the links below) I needed to collate again and add the recent Twitter exchange with Jonathan Haidt (also below):

Vive la Différence – April 2015 prompted by gender difference denial by Alice Roberts with Michael Mosely in BBC2 Horizon 29th September 2014. Gender difference is a good thing, better not to deny it.

Vive la Différence – Let’s Get Physical – March 2016 prompted by PC response to Djokovic opinions about earnings of female tennis players. (Maybe need to also to consider the recent BBC presenter earnings gender differential – more market than either psychological or physical gender differences – where market is probably most appropriate for celebrity media presenters – so long as BBC is expected to behave as a market player, but that’s a separate non-gender issue?) 

Vive La Différance, Again – Feb 2017 prompted my more general polarisation issues in our “Age of Extremes”. Gender is obviously binary (conventionally, there are of course grey, fuzzy exceptions) but true of so many this<>that labels.

The Google Gender & Diversity Furore – August 2017 my original reaction to the Google memo.

The reason to collect these previous links, a series of tweets via Jonathan Haidt regarding The Economist analysis of the Google Memo.

That is the reasoning and analysis are indeed thorough and nuanced, but the actual argument is too biased – clearly intended to attack the offender for his motives, rather than stick to the facts.

“gender balance as the engine
of intellectual diversity and innovation”

=====

[Post Note: An earlier recent “Brains and Gender” addition, and the biological never ends:

Excellent piece by Kenan Malik (originally from last weekend’s Observer).

“words still fail us when we talk about muslims”

It’s what I call “identity politics“, [and more here] but the choice of words is invariably polemic or politically motivated rhetoric rather than objective reporting of facts. These are subjectively entangled issues, and the lazy solution is the familiar PC option of euphemistically avoiding the “offending” terms, but as Malik says we need to talk about them and we need to do it using language that recognises reality.

“trapped between hostility towards Muslims
& fear of creating such hostility
or of offending Muslims”

“a broader confusion about
the relationship between race and culture
[and religion and labelling]”

As some of the Twitter responses indicate, even when being careful, each use of each chosen word nevertheless reduces individual sentences to this <> not-this duality. This is inevitable if we are to consider each point, each distinction, one at a time. To reduce language to nebulous catch-all terms to talk about the whole, is to talk honestly about nothing. The trap of being caught between the Scylla and Charybdis of the PC and the Polemic.

Somewhere on the shelf behind me, I do have Critique of Pure Reason but I was pretty inexperienced in philosophy, 15 years or so ago, when I first (and last) tried to read it. I didn’t get very far. Which doesn’t mean I don’t appreciate Kant’s significance, just that I’ve formed my views on Kant from only second-hand readings (so far).

In “Five Books” Nigel Warburton interviews Adrian Moore on his 5 recommended Kant readings, four of which are Kant’s own writings. A seriously heavy-weight recommendation, but fortunately the interview style teases out some summary content.

It’s a good piece, recommended generally, but I wanted to capture this topical point:

We have knowledge only of phenomena – ‘phenomena’ is Kant’s word for appearances – and we [can never have any] knowledge of noumena … how things are in themselves.

Everything is completely causally determined in the phenomenal world. So how can there be freedom in the phenomenal world? The answer to that question is: there can’t be.

We have to regard our belief in our own freedom – [in our real noumenal selves] – as an article of faith.

I subscribe to the phenomenal / noumenal distinction – the world beyond our “experience” can never be “known”. Even though that boundary gets pushed back by prosthetic extensions to our experience of phenomenal properties all we are ever doing is building a better “model” of the noumenal according to the phenomenal and our Reasoning.

What bothers me is that causal determination is based on a Newtonian billiard-balls model, and the article of faith is a doubly convenient way to preserve faith in divine will too.

Whatever the value of what Kant has to say about limits to Reason – logical, ethical, categorical, the lot – what he has to say about free-will really cannot be taken seriously?

Two things I need to write about.

Jared Diamond on Common Sense, in Edge 2017. “common sense should be invoked more often in scientific discussions, where it is sometimes deficient and scorned” Too right.

And “Why do we use reason to reach nonsensical conclusions?” from the New Humanist which I’ve been meaning to respond to since April 2017.

Closed-minded adherence to the technicalities and process of rational – would-be scientific – discourse, is often at the expense of sensible content. Don’t throw your intuitions out just yet.