Maverick Philosopher

Malcolm Pollack on AI and its Threat: Determinism, Predictability, Free Will

Our friend Malcolm Pollock in Brake Failure expresses a reasoned pessimism about our future under AI. I share his concerns. Will we humans have a future? Or are we facing what I have elsewhere called the Ultimate Replacement?

In Stephen Wolfram on AI and Irreducible Complexity, Malcolm explains

. . . a distinction between two [disjoint] subsets of deterministic systems: those whose behavior are describable by simplifying formulas that can be used, by taking their initial conditions as inputs, to predict their future state, and those for which no such reduction is possible.

An example of the former is the movement of two bodies under mutual gravitational attraction, such as a planet and its moon, or the earth and a ballistic projectile. Given the masses of the two, and their initial positions and velocities, it is possible to calculate their positions for any future time.

A good example of the latter is what Wolfram examined at length in his book A New Kind Of Science (which I labored through when it cam[e] out in 2002): the behavior of “cellular automata“, simple systems whose behavior is defined by a small set of rules, but for which, given the system’s state at time t, the only way of determining its precise configuration at time t+n is actually to iterate over every step between t and t+ n. Chaotic systems, such as weather and turbulent flow, are of this kind. So is biological evolution.

Fascinating. What struck me is that the first type of deterministic system allows for prediction, whereas the second type decouples determinism from predictability. I would add that if time is a continuum, then there are continuum-many iterations between t and t + n, which implies then there will have to be continuum-many iterations total.  That would be the ultimate nail in the coffin of predictability, a nail that not even the ultimate claw hammer could remove.

A couple of further questions occur to me.

In the second type of determinism, what becomes of the distinction between determinism and indeterminism? There would presumably still be the distinction, but how could one tell  if a type-2 system was deterministic or indeterministic? Malcolm, glossing Wolfram and Greene, writes, "There is no quicker way, no shortcut, for predicting the future state of such systems than simply letting them run, and seeing what they do." That boils down to saying  that in the second type of deterministic systems there would be no way at all of predicting future states of such systems.  How then could one 'determine' (come to know) whether such a system was deterministic or indeterministic?

If the deterministic systems that really interest us are of the second type, then Laplace's Demon is, if not out of a job, then bound to be underemployed.

Second question.  If we humans are deterministic systems of the second type, might this permit a deterministic reduction of the much-vaunted free will that we feel ourselves to possess? I don't think so, but knowing Malcolm, he may want to take this ball and run with it.

Dennett on the Consequence Argument against Compatibilism

Substack latest

Victor Reppert on Soft Determinism

Victor writes,

Soft determinism is still determinism. And it's really not a different type of determinism. It is, rather, drawing different conclusions from determinism, or rather, not drawing the conclusion that we are not free and not morally responsible for our actions.

Exactly right.

Caesar, the Rubicon, Tenseless Truth, Determinism, and Fatalism

In a post the point of which was merely to underscore the difference between absolute and necessary truth, I wrote, somewhat incautiously:

Let our example be the proposition p expressed by 'Julius Caesar crossed the Rubicon in 44 B.C.' Given that p is true, it is true in all actual circumstances. That is, its truth-value does not vary from time to time, place to place, person to person, or relative to any other parameter in the actual world. P is true now, was true yesterday, and will be true tomorrow. P is true in Los Angeles, in Bangkok, and on Alpha Centauri. It is true whether Joe Blow affirms it, denies it, or has never even thought about it. And what goes for Blow goes for Jane Schmoe.

As a couple of astute readers have pointed out, the usual date given for Caesar's crossing of the river Rubicon  is January 10, 49 B.C. and not 44 B. C. as stated above.  If only the detection and correction of philosophical erors were as easy as this!

The erudite proprietor of Finem Respicem, who calls herself 'Equity Private' and describes herself as a "Armchair Philosophy Fangirl and Failed Theoretical Physicist Turned Finance Troublemaker," writes, "Caesar crossed the Rubicon on January 10, 49 B.C., reportedly (though perhaps fancifully) prompting Gaius Suetonius Tranquillus to comment Alea iacta est ('The die is cast.')"  And Philoponus the Erudite has this to say: 

I'm not sure whether you are deliberately testing the faithful readers of The Maverick, but the accepted date for Caesar and Legio XIII Gem. wading across fl. Rubico is 49 BCE, on or about Jan 10th. That's what is inferred from Suetonius' acct of Divus Caesar at the beginning of De Vita Caesarum (written 160 years after the fact) and some other latter sources like Plutarch.

So I stand corrected on the factual point.  Both correspondents go on to raise philosophical points.  I have space to respond to only one of them.

Equity Private asks, concerning the proposition expressed by 'Caesar crosses the Rubicon in 49 B.C.,' "But is it true in 50 BC?  In a deterministic universe, I think it is. In a non-deterministic universe I think it isn't. Are you a determinist?"

To discuss this properly we need to back up a bit.   I distinguish declarative sentences from the propositions they are used to express, and in the post in question I was construing propositions along the lines of Gottlob Frege's Gedanken.  Accordingly, a  proposition is the sense of a context-free declarative sentence.  A context-free sentence is one from which all indexical elements have been extruded, including verb tenses.  Propositions so construed are a species of abstract object.  This will elicit howls of outrage from some, but it is a view that is quite defensible.  If you accept this (and if you don't I will ask what your theory of the proposition is), then the proposition expressed by 'Caesar crosses the Rubicon in 49 B.C.' exists at all times and is true at all times.  (Bear in mind that, given the extrusion of all indexical elements, including verb tenses, the occurrence of 'crosses' is not present-tensed but tenseless.)  From this it follows that the truth-value of the proposition does not vary with one's temporal perspective.  So, to answer my correspondent's question, the proposition is true in 50 B.C. and is thus true before the fateful crossing occurred!

I am assuming both Bivalence and Excluded Middle.    Bivalence says that there are exactly two truth-values, true and false, as opposed to three or more.  If Bivalence holds, then 'not true' is logically equivalent to 'false.'  Excluded Middle says that, for every proposition p, either p is true or it is not the case that p is true.   Note that Bivalence and Excluded Middle are not the same. Suppose that Bivalence is false and that there are three truth-values. It could still be the case that every proposition is either true or not true. (In a 3-valued logic, 'not true' is not the same as 'false.') So Excluded Middle does not entail Bivalence. Therefore Excluded Middle is not the same as Bivalence.  Bivalence does, however, entail Excluded Middle.

Here is a simpler and more direct way to answer my correspondent's question.  Suppose some prescient Roman utters in 50 B.C. the Latin equivalent of 'Julius Caesar will cross the Rubicon next year.'  Given Bivalence and Excluded Middle, what the Roman says is either true, or if not true, then false.  Given that Caesar did cross in 49 B.C., what the prescient Roman said was true.  Hence it was true before the crossing occurred.

Let's now consider how this relates to the determinism question.   Determinism is the view that whatever happens in nature is determined by antecedent causal conditions under the aegis of the laws of nature. Equivalently, past facts, together with the laws of nature, entail all future facts. It follows that facts before one's birth, via the laws of nature, necessitate what one does now. The necessitation here is conditional, not absolute.  It is conditional upon the laws of nature (which might have been otherwise) and the prior causal conditions (which might have been otherwise).

If determinism is true, then Caesar could not have done otherwise  than cross the Rubicon when he  did given the (logically contingent) laws of nature and the (logically contingent) conditions antecedent to his crossing.  If determinism is not true, then the laws plus the prior causal conditions did not necessitate his crossing.  Equity Private says  that the Caesar proposition is not true in 50 B.C. in a non-deterministic universe.  But I don't think this is right.  For there are at least two other ways the proposition might be true before the crossing occurred, two other ways which reflect two other forms of determination.  Besides causal determination (determination via the laws of nature and the antecedent causal conditions), there is also theological determination (determination via divine foreknowledge) and logical determination (determination via the law of excluded middle in conjunction with a certain view of propositions).  Logical determinism is called fatalism.  (See the earlier post on the difference between determinism and fatalism.) 

Someone who is both a fatalist and an indeterminist could easily hold that the Caesar proposition is true at times before the crossing.  Equity Private asked whether I am a determinist.  She should have asked me whether I am a fatalist.  For it looks as if I have supplied the materials for a fatalist argument. Here is a quick and dirty version of an ancient argument known as 'the idle argument' or 'the lazy argument':

1. Either I will be killed tomorrow or I will not. 
2. If I will be killed, I will be killed no matter what precautions I take.
3. If I will not be killed, then I will be killed no matter what precautions I neglect.
Therefore
4. It is pointless to take precautions.

This certainly smacks of sophistry!  But where exactly does the argument go wrong?  The first premise is an instance of LEM on the assumption of Bivalence.  (2) looks to be a tautology of the form p --> (q -->p), and (3) appears to be a tautology of the form ~p -->(q -->~p).  Or think of it this way.  If it is true that I will killed tomorrow, then this is true regardless of what other propositions are true.  And similarly for (3).

Some will say that the mistake is to think that LEM applies to propositions about future events:  in advance of an event's occurrence it is neither true nor not true that it will occur.  This way out is problematic, however.  'JFK was assassinated in 1963' is true now.  How then can the prediction, made in 1962, 'JFK will be assassinated in 1963,' lack a truth-value?  Had someone made that prediction in 1962, he would have made a true prediction, not a prediction lacking a truth-value.  Indeed, the past-tensed and the future tensed sentences express the same proposition, a proposition that could be put using the tenseless sentence 'JFK is assassinated in 1963.'  Of course, no one could know in 1962 the truth-value of this proposition, but that is not to say that it did not have a truth-value in 1962.  Don't confuse the knowledge of truth with truth. 

Suppose I predict today that such-and-such will happen next year, and what I predict comes to pass.  You would say to me, "You were right!"  You would not say to me, "What you predicted has acquired the truth-value, true."  I can be proven right in my prediction only if I was right, i.e., only if my prediction was true in advance of the event's occurrence.

So the facile restriction of LEM to present and past is a dubious move.  And yet the 'lazy argument' is surely invalid!   

What is Fatalism? How Does it Differ from Determinism?

Robert Kane (A Contemporary Introduction to Free Will, Oxford 2005, p. 19) rightly bids us not confuse determinism with fatalism:

     This is one of the most common confusions in free will debates.
     Fatalism is the view that whatever is going to happen, is going to
     happen, no matter what we do. Determinism alone does not imply such
     a consequence. What we decide and what we do would make a
     difference in how things turn out -- often an enormous difference
     -- even if determinism should be true.

Although it is true that determinism ought not be confused with fatalism, Kane here presents an uncharitable definition of 'fatalism.' No sophisticated contemporary defender of fatalism would recognize his position in this definition. Indeed, as Richard Taylor points out in a well-known discussion (Metaphysics, Ch. 6), it is logically incoherent  to suppose that what will happen will happen no matter what. If I am fated to die in a car crash, then I am fated to die in that manner --  but it is absurd to append 'no matter what I do.' For I cannot die in a car crash if I flee to a Tibetan monastery and swear off automobiles.  There are certain things I must do if I am to die in a car crash.  As Taylor says,
 
     The expression 'no matter what,' by means of which some
     philosophers have sought an easy and even childish refutation of
     fatalism, is accordingly highly inappropriate in any description of
     the fatalist conviction. (Metaphysics, 3rd ed., p. 57)

Kane's contrast is therefore bogus: no sophisticated contemporary is a fatalist in Kane's sense. Should we conclude that fatalism and determinism are the same? No. I suggest we adopt Peter van Inwagen's definition: "Fatalism . . . is the thesis that that it is a logical or conceptual truth that no one is able to act otherwise than he in fact does; that the very idea of an agent to whom alternative courses of   action are open is self-contradictory." (An Essay on Free Will, p. 23.)

As I understand the matter, fatalism differs from determinism since the determinist does not say that it is a logical or conceptual truth that no one is able to act otherwise than he in fact does. What the determinist says is that the actual past together with the actual laws of nature render nomologically possible only one future. The determinist must therefore deny that the future is open. But his claim is not that it is logically self-contradictory that the future be open, but only that it is not open given the facts of the past, which are logically contingent, together with the laws of nature, which are also logically contingent.

Perhaps we can focus the difference as follows. Suppose A is a logically contingent action of mine, the action, say, of phoning Harry. Suppose I perform A. Both fatalist and determinist say that I  could not have done otherwise. They agree that my doing A is necessitated. But they disagree about the source of the necessitation.

The fatalist holds that the source is logical: the Law of Excluded Middle together with a certain view of truth and of propositions. The determinist holds that the source is the contingent laws of nature
together with the contingent actual past.