Dagens ord


Ansvar väger tyngre än frihet - Responsibility trumps liberty

14 juni 2020

More by Moffett: Nationalists and Patriots

About fifteen years ago, as my friend and I walked across the town square, I remember saying:

There's really only one Big Question remaining: People seem to fall into two categories - liberals and conservatives. Why is that? Is it inevitable?

This split seems to me to be the biggest, possibly the only, real obstacle to genuine progress. How should it be addressed? Can it be dissolved?

Since then, I've read a lot. And a lot has happened. If I was asked yesterday to write an essay to answer my own question, it would hopefully have started something like this.

People's outlooks on many pressing social issues betray how these roles [protection vs. provision] are valued differently depending on whether individuals subscribe to patriotism or nationalism. As most psychologists use the words today, these are habits of thought that represent distinct expressions of how people identify with their society. Sometimes lumped together, patriotism and nationalism become plain, and clash with each other, in troubled times. Depending on the person, perspectives shift to more nationalist or more patriotic viewpoints during periods of stress. Yet each individual usually sticks within a narrow range of attitudes over the course of his or her life; the sentiments emerge in childhood under the dual influence of inheritance and upbringing. 
The fundamental difference between nationalism and patriotism is that while individuals with both outlooks are devoted to their society, they relate to it differently. Patriots display pride in their people and a sense of shared identity and particularly of belonging; such a feeling comes naturally to those born in a country but can be acquired by immigrants. With most of their passion directed at their own group, patriots prioritize the needs of its members: making sure they have food, housing, an education, and so on. Nationalists have similar emotions but couch their identity in glorification. Their pride connects with prejudice. As obsessed as patriots can be with caring for the members, nationalists are absorbed with preserving a superior way of life by keeping the society safe and sound and putting their own people prominently on the world stage. 
Where it gets interesting is that patriots and nationalists have divergent ideas of who constitutes "their own people." Indeed, among the aspects of their identity nationalists admire are those that set the trusted majority apart. It's this position they guard. The extreme nationalist ardently protects each detail of that identity to keep the nation firmly associated with the angels. The priorities of nationalists include staunch demonstrations of loyalty, accepting customary rules of order, obeying leaders whom they see as responsible, and maintaining the established social relationships, most clearly between ethnicities and races. All of these values came to the fore as people settled down and began dominating others. Tradition-driven nationalists believe in their country no matter what. They commit to the status quo, at times at odds with those democratic ideals that allow for transformation: their personalities are less open to new experiences and social change. Compare this my country right or wrong stance to the outlook of patriots, who likewise give their country a high standing yet believe it must be earned rather than fought for, allowing that there are possibilities for improvement. 
In their attention to differences between groups, nationalists treat both people of other nations and minority citizens as outsiders, taking a narrow view of who is, at the heart, truly part of the society. They're more comfortable with the majoritarian idea of democracy in which the dominant people should have the primary say in governance. Their perspectives on moral and legal issues reflects this. I believe it fair to say that to a nationalist, a person of another ethnicity, citizen or not, is relatively more foreign
Earlier I called ants extreme nationalists because they stick tight to their colony marker - its scent - as a stamp of their identity. Indeed, though in our species a patriot can become as teary-eyed as any nationalist in displays of allegiance to a flag or anthem, nationalists are supersensitive to those symbols. For them brief exposure to a flag or an idolized leader incites an intense reaction - as does the absence of such an emblem when one is expected. Thus the uproar about gymnast Gabby Douglas not placing a hand over her heart while the American national anthem played in the 2012 Olympics, a lapse that to a nationalist made her gold-medal win too much about herself and not about the United States. The reaction was a sign of the sentiment that societies are entities: people don't compete in the games, countries do. 
Both the nationalist and the patriot perspectives can be logically consistent, with nationalists being more risk averse and on guard against anything that may contaminate their culture. They prefer to err on the side of separatism, erecting boundaries that might alienate those whose interests could differ from their own, while patriots are more sympathetic to opportunities for trade and cooperation with outsiders. 
In short, the nationalist is suspicious of diversity, while patriots often welcome it. Or at least they tolerate it, because even a patriot, no matter how equality-minded, isn't immune to prejudice: the ardor that patriots reserve for fellow society members of their own race or ethnicity still leads to discrimination as they subtly, and unwittingly, treat those like themselves more fairly. 
Why did these differences in patriotic and nationalistic attitudes evolve? The fact is that a clash in perspectives within societies, although at times so extreme as to verge on the dysfunctional, may have always been integral to human survival. Our varied expression of social viewpoints probably connects back to "timeless social concerns," as one research team put it. Each outlook is beneficial in certain contexts. This dimension of our social identity may be an adaption to balancing the needs for protecting and provisioning the society. Even though people with opposing perspectives might not see eye to eye, a society with too few or too many individuals at either end of the spectrum could be open to catastrophes. This promotion of behavioral diversity has parallells in unlikely animal species. Social spiders are most successful when their colonies contain both individuals that retreat from danger but fastidiously tend the nest, and bold ones that put more effort into defense against social parasites, which steal the colony's food; the colonies of certain ant species function most efficiently when they contain a similarly effective mix of personality types. 
For humans, the hazards of a population overly committed to either the nationalist or patriot extreme are manifest. Nationalists see the patriot's greater openness to weak borders and sharing across ethnicities as promoting social dependence and cheating, fears that reflect the competitive nature of groups present across species. Meanwhile, the prevalence of nationalists, convinced their ways are right and prepared to fight for them, means the dangers that nationalists fear can indeed be realized. Still, by readily espousing oppression and aggression, extreme nationalists bring to mind the historian Henry Adam's description of politics as a systematic organization of hatreds. Their outlook feeds on certain facets of psychology. It's intoxicating to fall in line against an enemy, at times at a whiff of trouble. For those swept up in a nationalist perspective, the swell of group emotions and awareness of common purpose gives life a greater meaning. Not just morale, but mental health improve among civilians when nations face conflict. The fact is, trigger-happy societies have long had an edge, with the impulse for war and the fear of attack critical in driving many social and technical innovations and the expansion of states. What's more, nationalists, adhering to a narrow interpretation about what behaviors are proper, have the advantage of being far mor tight-knit and homogeneous than patriots and better able to act together. All this is to say that the patriot's vantage point is and always will be a more onerous path. 
Because of the partiality for their group, displayed by patriots and nationalists in different ways, the troubles our societies face go deep. It's bad enough that a wicked act by one minority person - the Florida nightclub shooting, for example - can set off outrage at an entire minority population. But mistreatment can carry over to ethnicities unconnected to the tragedy. That's an outcome of how stereotypes strip away detailed understanding, making it easy to conflate groups to the point of creating such fuzzy and nonsensical categories as "brown people." Even when no conflation exists, prejudices can be linked, with the denigration of one people associated with the devaluation of others. Persons who fear for their safety, jobs, or way of life indiscriminately lump them together much as ancient societies did with the "barbarians" beyond their borders. The impulse is so strong that when a sample of Americans was asked what they thought of Wisians, nearly 40 percent regarded them poorly and did not want them as neighbors, even though they could have known nothing about them since the researcher had made the name up. 
Societies contain ethnicities and races that stick together despite the members' prejudices about each other. The usual view, voiced by William Sumner more than a century ago, is that friction with outsiders draws a society together. Clearly, that's not always true. The external forces that promote civil peace primarily galvanize the dominant people while often straining their ties to a society's other ethnicities when those groups are regarded as part of the problem. This tension among the members can cause a kind of social autoimmune disease, turning a society against itself. For all these tribulations, we may reasonably ask whether societies are necessary at all.

--- Moffett, M.W. (2019), The Human Swarm, pp. 340-343


I am eager to read the concluding chapters of Moffett's book. Twenty or so pages remaining.

Just for the record: The fact that 40 percent of Americans shun Wisians isn't primarily attributable to individual nationalist tendencies, but to ignorance - which in turn isn't primarily attributable to individual personality traits, but to engineered anti-enlightenment. Any reasonable system of governance would, over time, not only steer the U.S. towards liberalism - in a broad sense - but would simultaneously narrow the divide between liberals and conservatives. (So conservatives would move more rapidly towards liberal values, even as the gap remains.)


13 juni 2020

Inventing foreigners

The disbanding of a society is a time of reinvention. Any reading of history suggests society breakups mirror the breakup of a marriage. When one can't turn back from a split, years of repressed opinions come pouring out that may express the opposite of what had been professed a month, if not the day, before. As pressures to conform to social norms shift, diminish, or vanish entirely, people on both sides gain the latitude to explore ways of interacting that had been out of favor or considered heretical. Previously unacceptable acts can leap to the forefront, helping each group distance itself from those who are now other, reimagined as outsiders, such that they come to appear ever more foreign.

The evidence indicates that many of the modifications of daughter societies - their character displacement, to borrow again a term from biology - occur in the initial years after they go their separate ways. Their newfound freedom of expression may be a reason why. That's when language - and no doubt many other, less studied aspects of identity - undergoes the fastest rate of change, before settling into a relative stasis thereafter. Indeed, distinctions between societies, often enough, are an outcome not of their ignorance of each other due to geographical separation, but of their awareness of and interaction with each other. This would be conspicuously true after societies split up. The opportunities for independent thought and invention afforded by a newly minted society, leading to a convergence of perceptions around themes the members can celebrate as their own, can make its formative years a golden age. For example, the Declaration of Independence and US Constitution remain the reference points that Americans turn to for guidance when questions about the nation's governance arise. Based on what is known about modifications in identity, I believe this would have been the case over the course of our evolution as it is now.

Yet there would be a deeper psychological impetus for a reworking of identity to bloom right after a division. The sense of being adrift, their fates severed from the meaning and purpose the larger society once provided, would heighten the urgency of the people's search for a strong identity, and essence, that stands apart. Moreover, their identification with each other must actually matter. Certain groups, such as people experiencing homelessness or those who are obese, may be marginalized but don't create societies with identities of their own. Neither do sick or disabled chimps or elephants, even when others treat them as outcasts. These outliers fail to bond since they do not see others with their condition in a good light. They lack what psychologists describe as positive distinctiveness.

Hence the insights of psychologists suggest that the members of a start-up society will toil to distinguish themselves favorably. To achieve this, they improvise cherished attributes or express old ones in a special way. The process is analogous to the development of traits that biologists studying the divergence of species call isolating mechanisms. Whatever commonalities remain with the other society can be denied or ignored. Like divorcees not on speaking terms, the societies can break off contact, which would mean any shared history would be eschewed or forgotten. Regardless, no matter how alike the newbie societies might seem to outside eyes, reunification would quickly be impossible.

From: Moffett, The Human Swarm, pp. 261-262

10 juni 2020

Cockroach

There was nothing in the near-infinite compendium of EU rules and trade protocols of the customs union that prevented a member state from reversing the circulation of its finances. That did not quite represent permission. Or did it? It was a defining principle of an open society that everything was lawful until there was a law against it. Beyond Europe’s eastern borders, in Russia, China and all the totalitarian states of the world, everything was illegal unless the state sanctioned it. In the corridors of the EU, no one had ever thought of excluding the reverse flow of money from acceptable practice because no one had ever heard of the idea. Even if someone had, it would have been difficult to define the legal or philosophical principles by which it should be illegal. An appeal to basics would not have helped. Everyone knew that in every single law of physics, except one, there was no logical reason why the phenomena described could not run backwards as well as forwards. The famous exception was the second law of thermodynamics. In that beautiful construct, time was bound to run in one direction only. The Reversalism was a special case of the second law and therefore in breach of it! Or was it? This question was hotly debated in the Strasbourg Parliament right up until the morning the members had to decamp to Brussels, as they frequently had to. By the time they had arrived and unpacked and enjoyed a decent lunch, everyone had lost the thread, even when a theoretical physicist came specially from the CERN laboratories to set everything straight in less than three hours with some interesting equations. Besides, the next day a further question arose. Would what the scientist said remain true if he’d said it in reverse?

— McEwan, The Cockroach

2 juni 2020

GPT-3: Filtrering eller resonemang?





Kilchers beskrivning känns uppenbart riktig, snudd på trivial. Att artikelförfattarna överdriver hur mycket, och vilken typ av ’resonerande’ som försiggår är förvånande och närmast pinsamt.

Visst finns det mycket som är imponerande, eller åtminstone anmärkningsvärt kraftfullt. (Wow-faktorn minskar ju i proportion till hur mycket tid och plats modellen och träningen har fått.)

Och visst kan det få stora, och eventuellt negativa konsekvenser om det blir riktigt lätt att generera hyfsat kontextkänsliga och åtminstone hyfsat ’unika’ texter vars artificiella upphov inte alltid lätt går att upptäcka.

Och visst kan man tänka sig att detta blir särskilt pikant om och när det kombineras med mer avancerat resonerande.

Men att påstå att den här mekanismen utför något mer än just det Kilcher säger - det verkar bara dumt.

Sen kan man ju alltid fråga sig var gränsen går mellan plagiat och ’försteåelse’, ’kunskap’, ’kompetens’... Om en elev/student får i uppgift att, säg, skriva en resonerande uppsats och lämnar in en kopia av en befintlig text, så är det ett uppenbart plagiat och inget bevis på egen förståelse. Om eleven läser och klipper och klistrar från flera olika texter på ämnet så är risken fortfarande stor att inlämningen flaggas i Urkund.

Men, bortsett från det verkligt originella, är det inte detta alla gör - och uppmanas att göra! - både för att lära sig och för att uppvisa bevis på att lärande har skett? Åtminstone om vi låter mängden lästa och ’plagierade’ texter och textsnuttar växa tillräckligt mycket. Och hur många uppsatser, skrivna ’med egna ord’, ärligt uppsåt och vad vi skulle erkänna som genuin förståelse, kunskap och kompetens, skulle godkännas om vi uppgraderade Urkund till GPT3? (Beroende på hur vi ställer in och tolkar granularitet och känslighet.)

Detta gäller förstås - hittills - bara i specifika typer av uppgifter, som egentligen inte kräver mer flexibilitet än vad GPT3 uppvisar. Men de är ju faktiskt väldigt vanliga i verkligheten. (Om än inte tillräckliga.)

Det mest intressanta var väl det om ’explicability’. Det verkar ju faktiskt vara ganska straightforward i just denna typ av fall. Och det går säkert att generalisera. Men troligen kommer förklaringarna att kännas mer otillfredsställande, ju svårare uppgiften hade varit för en människa att, i princip, utföra. Förklaringen finns där, men (den oreducerbara) komplexiteten och/eller den upplevda irrelevansen (i förhållande till mänskliga kognitiva begränsningar eller tankekulturer) kommer troligtvis att motverka den nytta som man hoppas få.

31 maj 2020

Intellectual asceticism and the collapse of Western civilization

"These so-called 'holistic' approaches still focused almost entirely on natural systems, omitting from consideration the social components. Yet in many cases the social components were the dominant system drivers. It was often said, for example, that climate change was caused by increased atmospheric concentrations of greenhouse gases. Scientists understood that those greenhouse gases were accumulating because of the activities of human beings: deforestation and fossil fuel combustion. Yet they rarely said that the cause was people and their patterns of conspicuous consumption.

"Other scholars have looked to the roots of Western natural science and religious institutions. Just as religious orders of prior centuries had demonstrated moral rigor through extreme practices asceticism in dress, lodging, behavior and food - in essence, practices of physical denial - so too did physical scientists of the 20th and 21st centuries attempt to demonstrate their intellectual rigor through practices of intellectual self-denial.

"These practices led scientists to demand an excessively stringent standard for accepting claims of any kind, even those involving imminent threats. In an almost childlike attempt to demarcate their practices from those of older explanatory traditions scientists felt it necessary to prove to themselves and the world how strict they were in their intellectual standards. Thus they places the burden of proof on novel claims, even empirical claims about phenomena that their theories predicted. This included claims about changes in the climate.

"Some scientists in the early 21st century, for example, had recognized that hurricanes were intensifying. This was consistent with the expectation, based on physical theory, that warmer sea surface temperatures in regions of cyclogenesis could - and likely would - drive either more hurricanes or more intense ones. However, they backed away from this conclusion under pressure from their scientific colleagues.

"Much of the argument surrounded the concept of statistical significance. Given what we now know about the dominance of nonlinear systems and the distribution of stochastic processes, the then-dominant notion of a 95 % confidence limit is hard to fathom. Yet, overwhelming evidence suggests that 20th century scientists believed that a claim could be accepted only if, by the standards of Fisherian statistics, the possibility that an observed event could have happened by chance was less than one in twenty.

"Many phenomena whose causal mechanisms were physically, chemically or biologically linked to warmer temperatures were dismissed as unproven because they did not adhere to the standards of demonstration.

"Historians have long argued about why this standard was accepted, given that it had neither epistemological nor substantive mathematical basis. We have come to understand the 95 % confidence limit as a social convention  rooted in scientists' desire to demonstrate their disciplinary severity.

"Western scientists built an intellectual culture based on the premise that it was worse to fool oneself into believing in something that did not exist than not to believe in something that did. Scientists referred to these positions  respectively as type 1 and type 2 errors, and established protocols designed to avoid type 1 errors at almost all costs. One scientist wrote: 'A type 1 error is often considered to be more serious and therefore more important to avoid than a type 2 error.' Another claimed that type 2 errors were not errors at all, just missed opportunities.

"So while the pattern of weather events was clearly changing many scientists insisted that these events could not yet be attributed with certainty to anthropogenic climate change. Even as lay citizens began to accept this link, the scientists who studied it did not. More important, political leaders came to believe that they had more time to act than they really did.

"The irony of these beliefs need not be dwelt on"

---

Transcribed from: Oreskes & Conway, "The Collapse of Western Civilization" (2014/2018), chapter 2 in the audio version, minutes 8-13.

23 maj 2020

The tribalism dogma

Given that, for significant numbers of humans, the First and Second Great Expansions have occurred, people who subscribe to the dogma that human beings have a tribalistic moral nature have two main options. Either they can give up the dogma entirely and admit that they simply didn't appreciate the flexibility of the moral mind because they confused the moral mind with the kind of moralities it was first expressed in; or they can soften the dogma by acknowledging that culture has stretched the tribalistic evolutionary leash in some cases, but assert that its doing so was a matter of going against our evolutionary grain, and that consequently any shift toward inclusion is bound to be anemic and unstable.

Obviously, I think the first response is the best: the belief that we are beings with a tribalistic moral nature should just be abandoned. To those who take the second option, my reply is simple: given that human moralities exhibit great diversity and that some people's moralities are not tribalistic, why should we say that our moral nature is tribalistic? It won't do for the die-hard defender of the dogma to cite evidence of the pervasiveness of tribalistic moral attitudes and beliefs, whether the evidence is historical or based on experiments, even if it is cross-cultural, because any such evidence is compatible with the hypothesis that our moral nature (the moral mind) is neither tribalistic nor inclusive, but rather so flexible as to be capable of being expressed in either tribalistic or inclusive moralities, depending on the environment.

In other words, to take historical or experimental evidence of tribalistic moral attitudes or behavior as conclusive confirmation of the thesis that humans have a tribalistic moral nature is to ignore the biased-sample problem I noted earlier. If the human-constructed niches in which the moral mind can get expressed in inclusive moralities are rare and recent, then concluding that human moral nature is tribalistic because most human moralities have been like that would be no more cogent than concluding that water flea nature includes protective spines and helmets because most of the water fleas you happened to have observed has those features.

The tribalism dogmatist has recourse to one last desperate fallback position: he can admit that human moral nature sometimes permits nontribalistic moralities, but insist that tribalism nevertheless is part of the moral mind itself, and that it is therefore still accurate to say that humans have a tribalistic moral nature. For the reasons given in the preceding paragraph, I don't think that we should say that tribalism is an element of our basic moral psychology, part of the moral mind itself - unless we quickly add that the moral mind also includes the capacity for inclusion. But if one grants that the moral mind is tribalistic in that weak sense - that tribalism is only one aspect of a moral nature that also encompasses inclusion, the result is a Pyrrhic victory.

Why? Because the thesis that humans have a tribalistic moral nature loses its punch if one admits that humans can act contrary to that aspect of their nature because of another aspect of their nature, namely, their capacity for inclusive morality. Admitting that our moral nature is dualistic saps the force of the assertion that we have a tribalistic moral nature.

Suppose that the last-ditch tribal dogmatist rejects the idea that our moral nature is dualistic and in response to the obvious fact that some moralities are now inclusive says that inclusion nonetheless goes against our nature. That is, suppose that he still asserts that our moral nature is tribalistic, period, not tribalistic and inclusive. Allowing for our moral nature to be overridden - saying, in effect, that we can act "unnaturally" - dilutes the notion of the nature of a thing. And that greatly reduces the interest and importance of the grand thesis that we are beings with a tribalistic moral nature. It also renders that thesis incapable of yielding any significant predictions about the scope of potential moral change or the space of possible moralities.

What I have to say in this book will still be important even if you dispense with talk of human moral nature altogether and fall back to a weaker, much less sexy thesis: namely, that human beings, by virtue of their evolutionary history, are predisposed to behave in a morally tribalistic way. To say that they are predisposed to tribalism means that exhibiting tribalistic moralities is the default position for humans, the way they tend to act, even if that tendency is sometimes not realized because of cultural influences. Its the default position, because the disposition to tribalism is an especially powerful aspect of our moral psychology.

If you hold that much weaker thesis, you can read me this way: I will show that in spite of this supposed predisposition, some humans have developed inclusive moralities, moralities that are not merely aspirational but socially and politically potent; and I will explain how they did that. In other words, I'll explain how they moved from situations in which that supposed predisposition largely determined the character of human moralities to situations in which that predisposition was inhibited or overridden or neutralized to such an extent that they developed inclusive moralities. In explaining this shift, I will show that the supposed predisposition to tribalism is not nearly as severe as one might think. I'll show that inclusive moralities are likely to persist, if the environments that are friendly to them are sustained. Inclusive moralities are fragile, in the sense that the capacity for tribalism never disappears; but that doesn't mean that inclusive moralities are inherently unstable and doomed to decay.

Finally, let me emphasize that there is a way of interpreting the experimental and historical evidence of tribalistic moral attitudes and behaviors that is compatible with rejecting the Tribalism Dogma, even in its weaker forms. Such evidence can be explained without concluding that humans have an especially powerful, deeply rooted, biologically based predisposition to tribalism if we make the following reasonable assumptions: (1) the moral mind is highly flexible; (2) most humans have lived in environments in which that flexibility issued in tribalistic moralities; (3) culture has developed in ways that sustain tribalistic moralities, even when they no longer promote reproductive fitness and are no longer necessary for successful cooperation; and (4) some people have an interest in maintaining (or resurrecting) tribalistic tendencies and they have the ability to do so effectively.

Each of these assumptions is highly credible. [...]

From: Allen Buchanan (2020),
Our Moral Fate: Evolution and the Escape from Tribalism,
MIT Press, Cambridge: MA.,
pp. 25-28



3 maj 2020

Ekofascism?

”Ekofascism” är ett märkligt skällsord. Man måste nog vara sociopat för att använda det.


Med terminologin i artikeln skulle jag kunna omformulera mitt ursprungliga utrop så här: Om man inte åtminstone är en prudentiell antropocentriker, så är man en sociopat. Och ska prudentiell betyda något meningsfullt så måste det förr eller senare innebära (accepterandet av eventuellt nödvändiga) kompromisser där enskilda människors intressen får stå tillbaka för helheten (inklusive alla andra människor, nu och i framtiden, samt deras miljö - oavsett hur mycket egenvärde man tillmäter denna miljö, och hur "naturlig" man insisterar på att den ska vara).

Callicotts holism, åtminstone i sin ursprungliga form, tarvade kritik, helt klart. Bland annat för att den inte var konsekvent. (Enskilda individer är ju en del av helheten och har därmed intrinsikalt värde.) Men Regans rop om "ekofascism" var minst lika missriktat. Den "misantropi" som han och andra varnar för låter i mina öron som ett brott mot hans egna principer: miniride och worst-off. Det verkar som om minsta inskränkning av en enskild individs handlingsutrymme är tabu för honom, och andra kritiker.

Diskussionen om "aristokrati" och "kolonialism" är i det här sammanhanget villospår. Varför skulle ex. uppmaningar till barnbegränsning etc. rikta sig i första hand till tredje världen; mer naturligt är väl att de i första hand riktar sig till väst och kompletteras med en uppfordran (plikt) att påskynda den sociala och teknologiska standarden i tredje världen, så att de så småningom kan följa efter - av upplyst egenintresse. Men det är kanske just där skon klämmer?

Om man vägrar att acceptera Malthus, och om man inte kan tänka sig en liberalism som modererar sina ambitioner, så förstår jag inte vad man har att göra inom miljöetik. Då har man ju på förhand deklarerat att man prioriterar någon princip högre än vår gemensamma långsiktiga överlevnad - och därmed i förlängningen även efterlevandet av denna princip. Om miljöetik ska vara intressant så måste den ju snarare handla om vilka principer som är långsiktigt hållbara.

Planet of the humans


Michael Moores nya film har fått mycket kritik. En del av den är korrekt (men av marginellt intresse); en del visar sig förhoppningsvis så småningom vara berättigad (vilket isåfall också berättigar Moores uppfordran att visa detta); en del är önsketänkande; en del körsbärsplockande (vilket Moore förvisso gör sig skyldig till, men bevisbördan ligger inte på honom); en del är positionering; en del i bästa fall välmenande politik; och en del är ren hysteri.

Men bara det faktum att McKibben, Sierra Club m.fl., och deras anhängare är så ömfotade visar att Moores inlägg är viktigt.

Påståendet att Moore går fossilindustrins ärenden, medvetet eller inte, är löjeväckande desperat. Dessutom osar det av dåligt samvete och whataboutism. (Huruvida Moore är oförvitlig är helt irrelevant.)

Sanna miljövänner välkomnar Moores ifrågasättande av teknikoptimism och termodynamiskt handviftande, och hans betoning av problemen med befolkningsökning, livsstil, tillväxt och kapitalism.

Kritik som inte explicit gör detta - och jag har inte sett någon sådan - tar jag inte på allvar.

---

Och där kom det:

25 ways of looking at AI



Nu är det äntligen dags. Det ska bli spännande att se vilka författare jag håller med, och vilka som snackar goja. En gissning:


TROLIGEN BRA

Lloyd (Nja... Försiktig, ej övertygande)
Pearl (Ja! Ska genast köpa hans bok)
Russell * (Ja! Mycket är bra)
Dennett (Jodå. Fortfarande spänstig, och börjar - liksom Dyson - att närma sig Deacon)
Tegmark (Jaaa!)
Tallinn (Jaa!)
Wolfram (Jaa. Cool!)


OSÄKRA

Brooks (Helt OK)
Wilczek (Cool!)
Griffiths (Intressant. Läser gärna mer. Men hur är det med the big picture?)
Dragan (OK)
Anderson (Haha! Kul)
Kaiser (Jag blir nyfiken, vill läsa mer)
Gershenfeld (Bra. Spännande)
Ramakrishnan (My man! Ska läsa mer. Men inte om intelligens - där är han för "blöt")
Pentland (Mkt intressant! Asimov, Tuschin m.fl... Mercier & Sperber skulle behöva läsa detta)
Obrist (Öhh... nja...näe)
Galison (Trevlig)
Jones (Bah!)


TROLIGEN DÅLIGA

Dyson * (Mmm... Tänkvärd och kanske viktig, men saknar konkretion)
Pinker (Woah! Det är något skumt med denna styvnackade fanatism...)
Deutsch (Far out! Way too far out för att vara relevant. Hur mycket stryk fick D i skolan?)
Hillis * (Himla bra! Visdom)
Gopnik (Bra! Kant! Men varför slutsatsen att vi inte behöver oroa oss?)
Church (Öhh... Skriv som folk - för folk! Orka!)


Jag uppdaterar det här inlägget efter hand. Intryck efter läsning inom parentes. 

11 apr. 2020

Human Compatible

1) Mycket är mycket bra. Tack till Olle Häggström för tipset.

2) Idén om att ersätta mål med osäkerhet är verkligen djup.

3) Det förvånar mig lite att Russell är så GOFAI-fokuserad.

4) Han skriver inte mycket om varför de tidigare GOFAI-initiativen ebbade ut.

5) Han motiverar inte heller varför, eller hur, de nu skulle kunna bli mer framgångsrika.

6) Det förvånar mig att han så kategoriskt avfärdar potentialen hos neurala nät.

7) Och att han kategoriskt avfärdar analogin mellan mänskliga hjärnor och neurala nät.

8) Det senare motiverar han inte.

9) Det förra motiverar han bl.a. med att upptäckt av och resonemang med abstrakta hierarkiska begrepp inte är möjligt.

10) Men samtidigt påvisar han, och sätter visst hopp till, att just olika begrepp faller ut av IRL.

11) Även sådana som vi själva inte har hittat eller kan hitta.

12) Hur hänger detta ihop med hans tvärsäkra uttalande att system måste vara förståeliga?

13) Sista delen, där huvudidén presenteras, inleds med tre principer. När jag först läste dem tänkte jag direkt: Det här är ju upplagt för wireheading - Harriets egen hedonistiska spiral, alltså. Goodharts lag är ju liksom det existentiella villkoret i ett nötskal. Om och när det blir för kort sträcka mellan mål och belöning så mister målet sitt värde.

14) Även om Russell då och då snuddar vid detta så tar han inte tag i det ordentligt.

15) Närmast kommer han på de sista sidorna där han talar om en allmän bekvämlighet, men jag tänker mer på individuella upplevelsemaskiner.

16) Överhuvudtaget känns kapitel 9 alldeles för fragmentariskt och kortfattat. Jag upplever inte att han tar problemen på allvar, eller att han behandlar dem tillräckligt noggrant. Det är ingenjören som talar.

17) Fast, visst, han är lösningsorienterad, och det är ju bättre att vara konstruktiv än att bara filosofera över svår- eller olösliga dilemman.

18) Och visst är grundidén god - även om jag är osäker på hur väl de tre principerna fungerar.

19) Men när det gäller aggregation så känns det återigen som om diskussionen är lite lättvindig: ”Sånt här får vi klura ut efter hand”.

20) Jag menar, de filosofiska problemen kan ju inte designas bort eller hanteras med ingenjörsskap.

21) Russell verkar vara skeptiskt till inte bara nudging och liknande, utan också till paternalism i någon form.

22) Jag tror inte att vi klarar oss utan sådan - vare sig med eller utan AI.

23) På ett ställe skriver han något i stil med att eftersom det inte är en bra idé att skapa helt lojala AI (till någon individ) så måste vi ”stoppa in lite altruistiska och egalitära algoritmer” i varje personlig assistent - men då får vi kanske istället problemet att de rymmer hemifrån och ansluter sig till EA-rörelsen... så liiite lojala måste de få vara ändå. Det här stycket (och texten omkring det) läste jag flera gånger. Men jag förstod inte riktigt hur han menar. Eller snarare, jag förstår inte hur det hänger ihop med texten i övrigt. Det känns som om detta är en (av flera) svaga punkter i hans resonemang.