ChatGPT and the Defecating Duck

For dog owners, the intelligence of dogs shows itself (among other things) in their ability to learn tricks. For cat owners, the intelligence of cats shows itself (among other things) in their disdain for learning tricks. 

When Alan Turing conceived of a way to tell computers and humans apart, now known as the Turing Test, he called it the Imitation Game. His first example was to ask a computer to write poetry - specifically a sonnet on the subject of the Forth Bridge. And his idea of a plausible answer for the computer was to say: Count me out on this one. I never could write poetry.

No doubt many people have tested ChatGPT with exactly the same question. When Jessica Riskin tried it, she was not impressed by its efforts. She found Turing’s imaginary machine’s answer (Turing imitating a machine imitating a human) infinitely more persuasive (as indicator of intelligence) than ChatGPT’s. Turing’s imagined intelligent machine gives off an unmistakable aura of individual personhood, even of charm.

An earlier article by Professor Riskin described a mechanical automaton that attracted large admiring crowds in 18th century Paris. This was a generative pretrained transformer in the shape of a duck, which appeared to convert pellets of food into pellets of excrement. The inventor is careful to say that he wants to show, not just a machine, but a process. But he is equally careful to say that this process is only a partial imitation.

Whereas ChatGPT's bad imitation of poetry is real shit.

---

Jessica Riskin, The Defecating Duck, or, The Ambiguous Origins of Artifical Life (Critical Enquiry, 2003)

Jessica Riskin, A Sort of Buzzing Inside My Head (New York Review of Books, 25 June 2023)

Alan Turing, Computing Machinery and Intelligence (Mind 1950)

Smart Guns

Just suppose that smart guns were safer than ordinary guns.

For example, if someone grabbed your gun and tried to point it at you. If it was a smart gun you'd be perfectly safe because there is a fool-proof mechanism that prevents its use by an unauthorized person.

As with any technological advance, some people are sceptical. How much do you trust new technology? Suppose the gun doesn't work when you need it. Maybe an electromagnetic pulse (triggered by terrorists or natural solar activity) might take out all weapons in the area. Or maybe the bad guys (or the FBI) can hack into this mechanism and disable your gun before they attack you.

Meanwhile, like many technological advances, there are political implications. In the USA, the key question is whether such a mechanism might help reduce gun violence. Some gun control activists think such a mechanism would be pretty irrelevant.

But that doesn't stop the gun rights activists freaking out at the prospect of any damn technology on their precious weaponry. A shop owner in the US claims to have received death threats from pro-gun lobbyists for offering to sell the weapons. Meanwhile, as Joseph Steinberg suggests, an obsession with smart guns may inhibit other technological innovations that could make guns and gun-owning safer.

The belief here is that once these smart guns are available, by a process of technological determinism, they will become irresistible to legislators. Before long, they fear, you won't be able to buy regular guns.

Obviously that's a cause worth killing for.

---

David Kopel, Brady Center lawsuit to use “smart” gun mandate to trigger handgun ban in New Jersey (Washington Post 22 May 2014)

Karen McVeigh, Gun control groups accuse New Jersey of ignoring 'smart gun' law (Guardian 21 May 2014)

Michael S. Rosenwald, Maryland dealer, under pressure from gun-rights activists, drops plan to sell smart gun (Washington Post, 1 May 2014)

Joseph Steinberg, Why You Should Be Concerned About The New 'Smart Guns' (Whether You Love Or Hate Guns) (Forbes 4 May 2014)

Nicholas Tufnell, Smart guns: How smart are they? (BBC News, 23 May 2014)

Eugene Volokh, Smart guns, electromagnetic pulse, and planning for unknown-probability dangers (Washington Post 23 May 2014)


See also Batman/Catwoman: Trail of the Gun (hat tip @ChBrain).

Related post Technological Determism (December 2020)

Asking Stupid Questions

A judge has criticized a jury for asking stupid questions. I am not a lawyer, but I've always understood that the jury is an essential component of our legal system, and I'm sure the judge will have been careful not to express his criticism in a form that could be interpreted as contempt.

But what counts as a stupid question?

1. Some say there are no stupid questions, only stupid answers. When Bill Gates asks "What is a network?", this could either be interpreted to mean that Bill Gates is stupid or alternatively that he is very clever. (Smart money goes with the second of these two possibilities.) See my post What's the difference between judges and geeks? (April 2010).

2. A common purpose of a stupid question is to prompt a useful answer or insight. Socrates used to ask dumb questions to make his pupils think more deeply about some subject - this technique is known as maieutic (from the Greek word for midwife).

3. Where a group is required to come to some judgement, such as a jury trial, a member of the group may request clarification on some point, not because she is personally unclear about this point, but because she believes some other member of the group may be unclear. (So for example, if one jury member kept talking about religious belief, another member might pose a question to the judge to confirm that religious belief was not relevant to the case.)

4. This particular case raises challenging legal questions that have not been tested in court for a very long time. Where the defence of marital coercion relies on a private and unrecorded exchange between husband and wife, the interpretation of such common phrases as "reasonable doubt" may become problematic, as does the fine line between inference and speculation. A stupid jury might not have been troubled by such ambiguities, and the judge clearly expected them not to be troubled. @davidallengreen (Jack of Kent) affirms that "we should not be shocked that a jury dares to ask basic questions; we should be concerned that juries do not ask basic questions more often."

5. If the judge offers an opinion about the jury's competence without being aware of the discussion and group dynamics that may have given rise to the jury's questions, this appears to be based on speculation rather than inference. It's a fine line, of course.

Financial Literacy

In which @CJFDillow reacts to @TimHarford's column on financial literacy.

---

In his article Illiteracy Rules, Tim quotes a survey where only 7% of Americans knew which of two financial offers was more favourable. We might reasonably worry that such ignorance (Tim calls it financial illiteracy) might leave some of the remaining 93% vulnerable to economic exploitation.

Tim discusses two possible remedies - financial education and regulation. He points out that both remedies have had limited success, so he thinks we probably need both.

In his post Financial Illiteracy - What's the Problem? Chris points out that financial education can be counter-productive.

"For one thing, I fear that the cure here - even if there is one - might be worse than the disease. In some cases, a little learning is indeed a dang’rous thing because our confidence sometimes rises faster than our knowledge. Lessons in financial literacy might therefore increase people’s willingness to take on debt, in the belief that they can handle it."

This suggests the need for something we might call second-order intelligence - being smart enough to avoid trying to be too smart.

Tim's survey compared two possible choices for a purchaser - payment by instalments, or taking out a loan and paying the full sum at the end. Financially literate Americans should be aware of two further choices - paying a smaller amount upfront and not getting into debt at all, or deferring the purchase until you've saved the money. At the interest rates quoted in the survey, both of the choices offered in the survey look almost equally bad. (In which case, why waste time calculating which is worse?)

Chris asks what the real problem is. Is it that people are financially illiterate, or financially irrational, or just financially impoverished. If it's about irrationality, then Chris thinks what matters is not so much financial planning as character planning, and he remarks

"The problem is, though, that it's not obvious that capitalism can thrive if people stick to these principles."

If capitalism benefits from weak-willed consumers, as Chris implies, then the survey reinforces the capitalist agenda by implying that there are only two choices (payment by instalments or loan) and ignoring the choices that would be selected by more strongly willed consumers. Tim tells us to ask ourselves why companies are so keen to let us pay in instalments, as if that were evidence that this is the most profitable for them. (This is not necessarily the case, as payment by instalments probably involves greater transaction costs.) But surely the real reason is not because they would make more money from payments by instalments, but because they are keen to pander to our irrationality and give us as many ways as possible to buy things we cannot afford.

False Intelligence

For those of us who don't (knowingly) have daily contact with the security forces, the idea of military/security intelligence generally only surfaces when there is a problem with it. Tony Blair had a little difficulty with a "dodgy dossier" not long ago; and now the White House is faced with similar political problems, apparently of its own making. In a blog posting The Lies that Lead to War, Rodger produces some historical parallels from the Vietnam War, and asks

Did the administration distort the evidence? Or, were the intelligence agencies completely and horribly wrong? A third possibility also exists: did some intelligence sources knowingly distort the evidence?

These questions rely on a central premise - that the purpose of intelligence is to deliver some kind of knowledge or truth. False intelligence then represents a betrayal of this purpose - either the relevant agency is incompetent (root cause: poor management, chronic underfunding and/or excessive reliance on surveillance technology?) or someone has distorted the evidence.

But this bureaucratic account of intelligence cannot be squared with the view from popular fiction, where spies dabble in grey half-truth and deception. So which is correct - the bureaucratic view or the fictional view? Must we choose between these two?

Perhaps it is the premise that is wrong. What intelligence does (and therefore in POSIWID terms its purpose) is not to produce knowledge or truth, but to support action. Those who wish to call the administration and/or the intelligence agencies to account are attempting to separate their criticism of the intelligence from their criticism of how the intelligence was used, but arguably this separation doesn't make sense. If the intelligence doesn't stand up, then the military action doesn't stand up, and vice versa.

In an adversarial legal system (such as the one in which Tony Blair qualified as a barrister) the purpose of evidence is to stand up in court. Justice is achieved by a kind of boxing match between two bodies of evidence - the winner is the one whose evidence stands up the longest. (The writer Adam Mars-Jones wrote an excellent long story Bathpool Park, based on a legal case presided over by his father, that illustrates the failure of this adversarial system to deal with the complexities of the case.)

Thus the purpose of intelligence is not to be true (what is Truth? asked jesting Pilate) but to stand up. Not quite the same thing.

WiFi Minefield

Bruce Schneier blogs about Wi-Fi Minefield (April 2005).

"Put aside arguments about the ethics and efficacy of landmines. Assume they exist and are being used. Given that, the question is whether radio-controlled landmines are better or worse than regular landmines."

Is an intelligent landmine better than a dumb one? At one level perhaps it is, because it might be more discriminating about whose legs it blows off.

But then military planners may judge that this technology makes it okay to have a greater number of more powerful landmines (on both ethical and efficacy grounds). Furthermore, it makes it harder for the good guys to clear landmines. So now the landmine situation has escalated.

As an anonymous comment to Bruce's post puts it:

"I don't think arguments about the ethics and efficacy of landmines *can* be put aside here, because surely the "smartness" of these mines will be used to justify laying more of them."

The problem here isn't the intelligent technology itself, it's the stupid (man-made) judgements that the intelligent technology supports.

A levelling law seems to operate in many systems, by which a local increase in intelligence is compensated by a loss of intelligence elsewhere. So what's the purpose of intelligence here?