Editor's pick: Holding out for a hero? Try a scientist
Sophia Chen describes the lack of consensus over interpretations of quantum mechanics, presenting it as some sort of crisis (7 January, p 11). But, despite that lack of consensus, quantum science has transformed the world in which we live. Many would hold that having a “true interpretation” is not required for science to drive progress.
But if quantum science is an example of a success, climate science is an example of a failure.
Climate science has failed because scientists and science journalists have failed to argue that, although scientific hypothesis is not fact, it is by far the closest we are going to get to fact.
Political leaders get elected by talking in absolutes. The one thing all traditional heroes have in common is that they can tell right from wrong: leaders aspire to that aura.
However, today’s real heroes do not champion what is “right”. They spend years painstakingly collecting and understanding data so that they can tell us clearly and concisely what is best. Today’s heroes are scientists.
Artificial intelligences could be set to be selfish
Sandy Ong raises fascinating questions about the ethics of autonomous cars (7 January, p 36). There is an even darker ethical issue not yet touched upon: programmed selfishness.
In a world where the vast majority of cars are autonomous and follow agreed rules, there will be a temptation for some car companies (or individual owners hacking their car's code) to deliberately program their vehicles to run red lights, poach parking spots, overtake dangerously, and otherwise drive more riskily, to get their passengers to their destinations sooner. They will be counting on other autonomous vehicles to behave selflessly and take evasive actions. If two such selfish cars meet, they may find themselves playing out the prisoner's dilemma in real time. By acting selfishly rather than cooperating, both cars will end up worse off.
And this is far from a fanciful risk. Compare the current automotive scandal in which some manufacturers went to great technological lengths to evade emissions regulations. They may have caused hundreds of premature deaths – for the sake of claiming better performance numbers than their competitors.
Perhaps automotive engineers should be studying not only philosophy, but also sociology.
Artificial intelligences could be set to be selfish
I can think of two solutions to the major moral dilemma facing autonomous cars. The first is for their owners to have general liability insurance, much as physicians must have cover for everything that might cause a patient injury or death.
Second, if the car is intelligent enough to recognise moral dilemmas, it should be smart enough to know or discover the owner's preference: do they always choose self-preservation, or kids first, or women and children first?
Insurance companies may then choose to set premiums based on the predicted risk of paying damages given what the car was programmed to choose.
Artificial intelligences could be set to be selfish
Driverless cars indeed raise moral and ethical issues. One that is rarely considered is that speed limits are set for human drivers. They are a compromise between going slowly enough to avoid an unacceptably high level of deaths, while going fast enough to get where we want without unnecessary delays.
We expect driverless cars to be much safer than those with human drivers. I modestly propose that speed limits be raised, to reduce delays while retaining the “acceptable” number of deaths and injuries.
First class post
What if we just haven't identified the smart ones as psychopaths?
Meryl Whisker a finding on an average psychopath's intelligence (28 January, p 12)
What we can learn from inscrutable translators
Nello Cristianini asks whether statistical machine translation is “teaching us something about how humans extract meaning from sentences” and answers “hardly” (26 November 2016, p 39). But surely, if all the old attempts at performing natural language translation using grammar-based sentence analysis have failed, and the statistical analysis of translated documents on the internet has been a resounding success, that may speak volumes about how natural language works.
Talking to machines may harm human protocol
Victoria Turk describes how people try to engage digital home assistants in social interaction (17/24/31 December 2016, p 16). Since they use what is normally a human-to-human channel, it is hardly surprising that we follow the normal protocols of communication, such as thanking the other party for assistance given. And it is good that we do so.
Which is the worse outcome: thanking a computer that cannot appreciate it, or failing to thank a friend who can, and will, miss it if omitted?
I am more worried by the possibility that in trying to emulate humans, through speech or any other human channel, computers will unintentionally train us to interact with their slightly alien presence better than with actual humans. This could damage communication between people.
Artificial intelligence knows no better
Allan Paxton suggests that “super-AI will surely inject a sorely needed mega-dose of rationalism into human affairs, since it will be unbound by (often psychotic) human bias” (Letters, 7 January). I beg to differ. Any super-AI will unavoidably have biases in the data on which it was trained built into it.
Atheism comes naturally to this reader at least
I can easily believe that I am not always rational, as Graham Lawton suggests while discussing innate superstition (10 December 2016, p 29). I could even justify it if I wanted: after all, it is hard work considering an action from all angles all the time.
I can believe that sometimes the “rational” thing to do is to take the shortcut, go with your gut instinct and spend the time you've saved lying in a hammock.
However, I am pretty sure that I, and many others I know, are atheist to the core. As such I disagree with Lawton's statement that “atheism is hard work” and “only skin deep”. It seems to me that belief in a god has to be taught, and regularly reinforced.
I suggest that atheism is the “natural” state for which no effort or education is required.
Rationality probably isn't a “natural” state, and it may be that we are not always as rational as we would like to think, but being rational is not the same as being an atheist. I can be a rational or an irrational atheist, but I don't see how you can be a rational theist.
Who are these 'people' who are showing bias?
You describe how people shown a picture of a black man were more likely to be biased if shown the picture during a heartbeat (21 January, p 7). Do you mean white people? Great care needs to be taken with the use of the word “people” when describing experiments. Apart from failing to help readers draw considered conclusions, it can come across as racist or otherwise prejudiced.
The editor writes:
• There's some evidence that black people can also show unconscious bias against black people but, examining the paper's methods, it turns out that the participants in this particular study were white men and women ().
Preserve habitat rather than resurrect species
Cloning species threatened by extinction sounds like a cute idea (17/24/31 December 2016, p 37). But which species would we choose to save? Who would decide whether to save, say, a rare plant or an iconic wild animal like the critically endangered rhino?
The biggest threat to most species is habitat loss, usually to meet human needs. It'd be far easier, cheaper and more sensible to curb the threat of extinction by preventing further destruction of the habitats needed for the survival of vulnerable species.
Warning: that's a half-boiled egg-quation
You give a formula for soft-boiling an egg (17/24/31 December 2016, p 86). I warn readers against using this, because it will frequently undercook the egg.
It fails to account for the volume of water in the pan and rate of energy supply. These factors are of crucial importance because adding the egg will typically take the water off the boil.
Untangling the cross over of visual fields
You say the “brain's right hemisphere… is the one that receives signals from the left eye” (14 January, p 10). Only the inner-side or “nasal” retinal optic fibres from each eye cross over at the , near where the optic nerves enter the brain. The right hemisphere receives information from the left visual field, rather than the whole of the left eye.
Siamese cats, and members of many species (including humans) with albinism, can have eye-to-hemisphere pathways that cross over almost completely, but for some reason tend to suffer visual problems as a result.
Inter-universal travel: I wouldn't start from here
Shannon Hall states that in alternative universes the laws of physics may differ more or less from ours: gravity might even work in the opposite direction (21 January, p 28). So is it possible that somebody in one of these alternate universes could develop a method of travel between them?