Intelligence 3: Ethics of Intelligence Psychology

What is The Bell Curve?

A controversial 1994 book advocating for public policies to protect high IQ individuals and limit reproduction among low IQ individuals.

Many claims share accurate data but have inaccurate interpretations

What are the key claims made in The Bell Curve?

• Intelligence is stable (with your age), genetically determined (inherent), and reliably measured (similar score every time you take).

• Intelligence tests are valid and unbiased.

• Intelligence is the best predictor of life outcomes.

• Welfare allows low IQ individuals to have more children, reducing societal fitness.

Why is the claim that "intelligence is determined by genetics" incorrect?

• It reflects genetic determinism, ignoring environmental interactions. → we cannot say if intelligence is more due to genes/how much does intelligence have to do with genes without accounting for all environmental interactions

• Parental IQ correlates with children’s IQ largely due to shared environments, not just genetics. → parents create environments that mirror their own (not genetical)

• Other cultural variables and environmental factors can increase/decrease test performance

• If we had all the exact same everything in environment, this claim could be measured but that isn’t possible in practice

What does "intelligence is highly predicted/correlated by genetic variation" really mean?

True (according to twin studies), but it reflects societal structures, not innate genetics (e.g., a biased system like allowing only certain groups privileges).

What’s wrong with attributing group differences in intelligence to genetics? (Lynn claim)

Environmental effects and gene-environment interactions are not equal across groups, so group differences are not solely genetic. → There is an infinite amount of variables and third variables including all environmental interactions and effects

Is intelligence the best predictor of life outcomes? (Jordan Peterson claim)

Partially true, but outcomes are influenced by third variables like family wealth, access to education, and healthcare.

Directionality problem → it is only a correlation rather than causation

• Life outcomes may be preserving your intelligence (healthcare)

• Third variables that can be very good predictors (family wealth)

• We cannot make causal conclusions about intelligence because we cannot manipulate it

What is the difference between narrow AI and general AI?

• Narrow AI: Solves specific problems (e.g., playing chess). → any kind of system

• General AI: Solves a wide range of problems, resembling human intelligence.

Why is narrow AI considered less intelligent than humans?

1. Learns slowly compared to humans even with more than a lifetime of human exposure to material

2. Learns differently (e.g., humans learn passively, not from constant feedback).

3. Makes mistakes humans wouldn’t. (different underlying processes)

4. Struggles with novel tasks outside its programming.

What problems exist with defining intelligence for AI and humans?

Tasks like chess or creativity tests may not accurately reflect intelligence; definitions may need revision.

Imagine taking an intelligence test, and discovering that all of the questions ask what

the capital city of a country is. This test is:

a) Potentially a valid test of intelligence

b) Potentially a valid test of crystallized intelligence

c) Potentially a valid test of fluid intelligence

d) Not a valid test of intelligence, because it tests general knowledge

b) potentially a valid test of crystallized intelligence → solving problems using existing knowledge

Not the best test for measuring crystallized intelligence but it’s still testing crystallized intelligence (not a good predictor but still potentially one)

[T/F] If we discover a population of people for whom variability in intelligence is 50%

accounted for by the environment, then the other 50% must be caused by genetic variability.

a) True

b) False

b) False → predicted vs caused

Traits are not determined by traits and environments, also interactions and third variables should be accounted for