Toward the Singularity via super-intelligence

George Rebane

Knowing my relationship with Dr Judea Pearl, fellow blogger Russ Steele sent me a link to a recent interview (here) on AI with Judea.  Longtime RR readers recall pieces citing my friend, colleague, and mentor Judea Pearl as the inventor of Bayes and causal nets, and one of the true pioneers of AI and winner of the Turing Award (here and here).  I was privileged to be one of his very small group of doctoral students contributing to his research and the writing of his landmark Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference (1988) introducing Bayes nets for which I was able to contribute my research in machine learning that wound up in what is known as the Rebane-Pearl algorithm (here and here).

Early on (late 1980s) our little group began to understand that the alternative ‘causal net’ that was also used in the literature for Bayes nets may not be appropriate.  To be frank, we were somewhat guilty of introducing that appellation ourselves before we recognized that our enthusiastic reach exceeded our grasp of causality as provided by our research.  In our paper on the Rebane-Pearl algorithm I introduced the concept of a ‘causal basin’ and showed (in my hand-drawn figure which somehow made it into the paper and Judea’s book) how the algo could take a dataset and divide it into several causal basins.  I recall having long discussions with Judea on the relationship of probabilistic inference and ‘true causality’ as we humans ascribe to the goings on in our environment.

A decade earlier Judea and I spent time during our family social get-togethers ensconcing ourselves with a bottle of wine to speculate on the future of intelligent machines and what computational, perceptual, and cognitive requirements they would need to achieve peerage with humans.  Over such discussions it became clear to both of us that machines would have to include the ability to handle the subjunctive and counter-factual ‘thoughts’, in addition being able to reason probabilistically a la Bayes nets.  It was Judea who came upon the idea that causality was a distinct cognitive beast that could not be tamed and understood through mere extensions of Bayes nets – a whole new calculus would be required.

Judea launched on that quest around 1990, and by the mid-90s had invented what today is known as the ‘do-calculus’. (here and here)  After more research Judea formed a small discussion group on advanced AI concepts including causality.  I was privileged to be invited as the only non-academic member of that group, which I recall included, among others, such luminaries as Leonard Kleinrock (leading modern queuing theorist and a true founder of the Internet) and Adnan Darwiche (a stellar Bayesian, Google Scholar, and former chair of UCLA’s Computer Science Department).  At that time I was heavily involved in the introduction of ecommerce with Bizrate.com, a pioneering start-up that became one of the early dot-com success stories.  Finally, from all this work, Judea wound up literally writing the book on causality – Causality (2000) and recently the more accessible The Book of Why: The New Science of Cause and Effect (2018).

With the do-calculus Judea introduced the concept of the ‘causal beam’ as the sequence of directed and mediating happenings that precede the caused event in question.  We can think of the causal beam as the one specific imperative sequence of nodes within a causal basin that is necessary for the event to occur.  In sum, such an approach now lets us reason with cause and effect.  Within my own concept of causality, the ‘actual cause’ of an event is the upstream progenitor event that probabilistically ‘seals the fate’ of the terminal caused event coming to pass.  There’s more to be said about that, but let’s get back to Judea’s interview.

What caught my eye about this interview was Judea’s assessment of the current state of AI research, especially its almost unseemly focus on what has come to be known as Deep Learning.  As Judea points out, there is much more to AI than Deep Learning, which is the resurrection on steroids of neural nets (NNs) which languished for decades due mainly to lack of storage and processing power to handle the large amounts of data needed for training a NN, and then having enough horses to compute such a net with many very large hidden layers of ‘neurons’.  Judea is a little harsh when he describes Deep Learning as basically being little more than “curve fitting”.

As I’ve described before, NNs take a large potful of data and discover certain very highly dimensioned subsets or sub-hyperspaces of it with either a supervised learning (deviation from target behavior used to update NN parameters) approach or unsupervised learning mode (no target behavior specified).  These subsets are delimited or bounded by highly dimensioned boundaries or discriminants that later allow a new data point to be identified as belonging to this rather than all the other bounded (discriminated) subspaces.  What makes such NNs valuable are the semantics associated with each of the subspaces – such as malignant tumor or simple cyst, Sally Smith vs everyone else, or turn right at 20 degrees/sec vs left at 10 degrees/sec vs stop.  And yes, one can think of the process of digging such discriminants out of the data as a form of fitting a curve to the data.  However deep learning NNs do a bit more since they can also discover the form or shape of the best discriminant rather than fit a no matter how highly-dimensioned ‘curve’ of a fixed functional form to the data.  And such a capability is an extremely important advance.

But Judea’s lament on today’s almost exclusive focus on deep learning is well supported.  Today AI advances revolve around the effort to make ever better associations – this data points to there being a ‘cat’.  But the super-intelligence barrier will not be penetrated until machines can also conceive of and deal with the ‘what ifs’.  In Judea’s words, “If we want machines to reason about interventions (“What if we ban cigarettes?”) and introspection (“What if I had finished high school?”), we must invoke causal models. Associations are not enough — and this is a mathematical fact, not opinion.”

The road to get there loops back to how machines can conceive reality; in other words how can they be gifted with a working ontology that allows them to model their current environment (like humans) and then deal with it and/or do things within it.  And going a bit further, machines must have ‘free will’ according Judea because evolution tells us that having the “sensation of free will serves some (critical) computational function.”

The evidence of machines actually having free will according to Judea is when “robots start communicating with each other counterfactually, like “You should have done better.”  The capacity of free will immediately invokes the notion of evil, which itself requires an operational definition.  According to Judea, evil occurs in an agent that has the capacity to believe and act on the belief that its “greed or grievance supersedes all standard norms of society.”  A robot is seen to commit evil when it ignores the maintenance of its known social norms in order to satisfy its greed or assuage its grievance.

There is more, but this should give all of us a good chunk to chew on in these pre-Singularity years, as we consider the new areas of research that computer scientists like Judea Pearl and other workers in machine intelligence are now contemplating that lie beyond the still marvelous and magical capabilities provided daily to machines through “curve fitting”.

Comments

[Barry Pruett]

Is it fair to compare the nature of man to what the possible nature of AI would be? A deep conversation re what even is the nature of man? What about the nature of AI?

[George Rebane]

BarryP 818am - Indeed it is. I don't know about your use of "fair", but such a comparison is mandatory as our biological life contemplates a future transit to 'silicon based' life. Many of us in the field believe that intelligence in biological form across the universe is a relatively short-lived phase of life - the enduring life forms of ET civilizations occur after they achieve their Singularity and ability to travel and build off-planet habitats.

[L]

George @ 2:29, that's awfully close to coming right out and saying it...

[George Rebane]

L 631pm - I didn't mean to be circumspect here Larry; what didn't I say that I should have?

[Bill Tozer]

This may or may not be on topic.

The Human Skills AI Can’t Replace

https://quillette.com/2019/09/25/the-human-skills-ai-cant-replace/

[Russell Steele]

– Advances in artificial intelligence raise new questions about the future of humanity. But what will AI mean for faith? And how could faith shape AI?

Queen Anne Lutheran Church in Seattle is exploring this topic in a series of forums under the theme, Digital Apocalypse: Will AI Destroy Us, with the next installment taking place this Sunday Sept. 29. The event features a talk by Dr. Ted Peters, Director of the Center for Theology and Natural Sciences in Berkeley, followed by a panel discussion that includes GeekWire’s Todd Bishop and church members Tim Porter of Madrona Venture Group and Aaron Boysen.

Rev. Dan Peterson, the church’s pastor, will round out the series on Sunday Oct. 6 with a talk on the “Faith of Droids,” exploring how the sense of mission and purpose in R2-D2 (Star Wars: A New Hope) and K-2SO (Rogue One: A Star Wars Story) could provide a model for faith to shape AI in this world. Carissa Schoenick of the Alan Institute for Artificial Intelligence spoke previously in the series.

[George Rebane]

RussS 808am - Great news to hear that some Christian churches are beginning to expand their understanding of life, especially wrt AI. Jo Ann and I taught a seminar on 'The Bible and Science' for a few years at a local church we belonged to. The topics discussed were definitely a difficult pill to swallow for some of the attendees. But IMHO as a Christian, in order to stay relevant and survive, Christianity will have to undergo a considerable expansion of its ontology.

[Russell Steele]

The Defense Department Plans to Build Radiation-proof CRISPR Soldiers

A Defense Department project plans to temporarily alter human genes, and shield people from deadly radiation exposure.

https://onezero.medium.com/the-government-aims-to-use-crispr-to-make-soldiers-radiation-proof-3e18b00c9553

I cannot find the blog post, but over a year ago, I wrote this was one direction humans could take to become galactic explores. In the past cosmic and gamma rays have modified earthlings genetic codes. And, will in the future during catastrophic events. The question is, who will be next Mother Nature or a skilled geneticist? Scientists have discovered microbes with genetic codes for surviving radiation. What role can AI play in the geneticist decisions to modify humans for galactic exploration?

[The Estonian Fox]

George @ 2:29 PM Sep 26.

If intelligent carbon-based life has a short lifetime, and then transitions to silicon-based life (SiLi), this question is still there - where are they? If they were inhabitants of planets in the first phase of star formations from 1 billion years (BY) to 4BY after the big-bang, they've had 10BY to propagate throughout the universe. Even at less than light speed, that seems like plenty of time to travel within an individual galaxy, if not the universe.

It may well be that we haven't tested enough living DNA to find SiLi; or perhaps SiLi can transition at will between Si-based and carbon-based life to 'trick' us now that we can perform DNA tests. But we haven't found any non-carbon-based life yet on earth. And using Occam's razor, the simplest explanation is that there is no SiLi here. And if not here, then not anywhere, because 'here' is indistinguishable from 'there', when limitless travel is possible. Therefore no interstellar travel by SiLi.

(If you meant off-planet habitats in the sense that SiLi would leave their home planet for other existing planets, then ignore the following sentences). And why build off-planet habitats, when so many billions of planet-habitats are available, free for the taking. And SiLi needn't require free oxygen since SiLi wouldn't necessarily use free O2 as an energy source. The most efficient energy source is the nearest star. As our own earth-resident green plants seem to know.

[George Rebane]

EFox 1232pm – Good points. Before responding, I’d like to recommend your reading a couple of previous posts, it would expedite the discussion. Please let me know.
https://rebaneruminations.typepad.com/rebanes_ruminations/2019/07/berkeley-seti-research-center-visit.html
https://rebaneruminations.typepad.com/rebanes_ruminations/2018/06/a-copernican-answers-fermi.html

RussS 1137am – We learn that damaging radiation is the kind that is energetic enough to ionize atoms. And the problem comes when the damaged atom is in a critical part of a living cell. A good discussion of this can be found here -
https://teachnuclear.ca/all-things-nuclear/radiation/biological-effects-of-radiation/effects-of-ionizing-radiation-on-dna/

From your cited article the description of how CRISPR will be used to “shield people from deadly radiation exposure is accomplished” is not made clear. CRISPR is a new gene editing tool that can only mess with our DNA, cutting out and splicing in new strings of base pairs. There is now way that CRISPR or anything else working within the body can stop radiation damage, that penetrating high energy photon will mangle any kind of molecule it encounters. They may be using ‘shield’ metaphorically in the sense that we will discover how to construct genes (with CRISPR) that will automatically work to repair any radiation damage that occurs to molecules. And that will require a catalog of such genes to be ‘inserted’ into a body to protect it since the are many kinds of cells and several types of damage (e.g. somatic and genetic) that radiation can inflict. Lots yet to learn.

Anecdotally, when on active duty I was detailed to go through the Army’s combat CBR school. Among the many things we were taught there was how to deploy troops that had received a highly likely lethal dose of radiation from a recent nuclear burst. The immediate aftereffect of such an exposure would be a short interval of good feeling, maybe even bordering on euphoria, as a goodly amount of the body’s energy resources were no longer consumed in routine cell maintenance and repair (because the cells had received a somatic shock the either killed them wholesale and/or made the remainder so injured that they no longer were able to carry out normal functions such as repairing themselves). So, for that short post-exposure interval the afflicted could actually have their mood elevated. In that event we were advised to consider deploying such troops in dangerous defensive positions or be the lead elements of a counter attack (tactical nuclear bursts were always to be followed by an offensive action of some sort).

[J. Barron]

George 2:25

Tardigrades are immune to high levels of radiation by bundling protective proteins around DNA. We have used the protein to protect genetically engineered (via CRISPR) human kidney cells. There is actually quite a bit of work going on in the field of Dsup.

To start: https://www.newscientist.com/article/2106468-worlds-hardiest-animal-has-evolved-radiation-shield-for-its-dna/

[J. Barron]

As an aside, for 7th grade science fair I boiled and froze and starved Tardigrades of oxygen by putting the in pure helium. I then "brought them back to life". They were never really dead.

[George Rebane]

JB 520pm - Thank you for that interesting info on Tardigrades. Reading about them years ago I didn't understand then and still don't today how any kind of bio-molecule can protect DNA from an ionizing photon. And the article throws little light on the matter - "It is mainly down to a bizarre protective protein they evolved that somehow shields their DNA from radiation damage. Short for “Damage suppressor”, Dsup appears to work by physically cuddling up to DNA and cocooning it from harm, but without disrupting its normal functions." Apparently "somehow shields" is the best that anyone knows, and that's not much of an explanation. Has that protein been isolated, reproduced, and wrapped around a detector exposed to a radiation beam to see what kind of attenuation it provides. And more importantly to then see the mechanism by which the protein absorbs the high energy photon, with what kind of damage to it.

[J. Barron]

George-
Yes, the protein has been isolated and replicated, and no, understanding is very incomplete regarding its mechanism(s) of attenuating radiation. I think as fascinating as the Dsup, is the ability of the tardigrade to almost instantaneously repair radiation-caused delamination of the DNA. Rumor in the industry suggests that the Israelis and Japanese are far ahead of us in Dsup evaluation and its ability to impregnate humans from radiation as well as pairing with stem cells to repair damage. It appears most of this study is defense based and not available to common eyes.

Now is a great time to find tardigrades. Pull off a golf ball sized chunk of greening moss in your yard from a tree or stone. Plunk it into a petri dish with water and put it under a dissecting microscope. With a little patience you will find them frolicking about. I love these guys.