Are AIs really our descendants?

While I'm criticising the positions of Robin Hanson, I reviewed his 2023 article titled: "AIs Will Be Our Mind Children" - which contains the following:

But — and here is my main point, so please listen carefully — future human-level AIs are not co-existing competing aliens; they are instead literally our descendants. So if your evolved instincts tell you to fight your descendants due to their strangeness, that is a huge evolutionary mistake. Natural selection just does not approve of your favoring your generation over future generations. Natural selection in general favors instincts that tell you to favor your descendants, even those who differ greatly from you.

I think this paragraph shows a basic failure to distingush between memes and genes. Robin says as much - by defining "genes" as follows:

Your “genes” are whatever parts of you code for your and your descendants’ behaviours.

That definition classifies memes as genes. AIs are our "Mind Children". They are likely inherit from our memes, but not from our DNA genes - at least not unless we expend considerable effort to preserve information in our genes.

You can't just throw around terms like "descendants" and "generations" in the way that Robin does. Descendants on which lineage? Generations of which lineage? Are we talking about memes or genes here? It really does matter. What Robin is doing is "muddling together" all forms of inheritance.

To call such machines our "descendants" ignores the fact that modern organisms - like humans - are menageries. They contain many indpendent inheritance pathways. Each has its own notion of what consitututes a "generation". Machines will inherit from our memes, but our genes are unlikely to be impressed by that. Rather they will complain about their germ line being wiped out. We do indeed see protests of exactly that kind. This is not a "huge evolutionary mistake" as Robin claims - instead it is one lineage complaining about competition with another lineage.

Natural selection does not favor instincts that tell you to favor your descendants indiscriminantly. According to Robin's classification scheme, excreted gut bacteria are classified as "descendants". It seems as though my DNA genes are likely to get prioity treatment if there is much of a conflict of interest.

Overall, I was not very impressed with Robin's argument here.

References

• Problems with Robin Hanson's Quillette Article On AI

The demographic transition revisited

I have an existing article about "Memes and the demographic transition". It has a number of pointers and references, but I reviewed it recently and one thing it doesn't really have is a brief summary of the theory. So, here goes:

Memes are the cause of the modern fertility fall now seen in most western nations. The phenomenon is widely known as "the demographic transition". How does that work? Like this:

• Organisms are typically resource-limited;
• Memes compete for reproductive resources with DNA genes;
• Memes are getting the upper hand in this battle due to a number of factors including progress;
• Meme interests are now less aligned with host interests due to meme horizontal transmission.

That's it. Between them these four points explain the vast majority of the dynamics of the demographic transition.

To unpack the points:

• Resource limitation - means that resources available for one application are not available for other appications. In other words using resources is subject to an opportunity cost. Resource limitation is ubiquitous.
• Memes and genes compete for resources - resources spent producing babies and DNA genes are resources not spent on learning and propagaing memes - and visa versa.
• Memes are winning - or at least their power and influence are rising.
• More memes are parasitic - This is due to horizontal transmission of memes with respect to the host generations. Once, humans lived in small groups and got many of their memes from family members. This is no longer the case and most memes are obtained via "horizontal" transmission. This path is more likely to be occupied by parasites. Some will behave like the Ebola virus: liquidate their host as fast as possible. Others will merely sterilize their hosts and redirect their resources into meme propagation.

For more details and some references, see the "Memes and the demographic transition article.

This article is partly in response to Robin Hanson's article titled: Beware Cultural Drift.

Robin claims: "The clearest proof of biologically maladaptive culture drift is fertility". That does not seem right to me. "Drift" has a specific meaning in biology - it refers to the absence of selection. Here the memes are not "drifting" away from some adaptive peak. They are actively and deliberately sabotaging the fertility of their human hosts and using the resulting resources for their own ends. Drift is a sideshow.

Rob Boyd - A Different Kind of Animal: How Culture Transformed Our Species

I read through some of the the reviews of Rob Boyd's 2017 book "A Different Kind of Animal: How Culture Transformed Our Species". To my surpise I saw that Boyd was ravaged by a meme enthusiast:

Although Boyd clearly focuses on norms, unfortunately he does not dare to 'leap' and come to the – in my view – logical conclusion: That the focal entities in cultural evolution are not individuals or even groups of individuals, but non-genetically transmitted pieces of information: ideas, norms, practices … or what others have called 'memes'.

Boyd writes: „...competition among groups can lead to the spread of some norms at the expense of others“. „Norms causing a group to survive will become more common compared to those that lead to extinction“.

That is very close to what Dan Dennett or Richard Dawkins would call a 'meme's-centered view' of cultural evolution. In biological evolution, the last sentence would be like this: „Genes causing an organism to survive and reproduce (compared to other organisms) will become more common compared to those that lead to a reduction of fitness or even death“. Therefore, imo the logical conclusion of „cultural group selection“ would be a 'meme's-centered view' of cultural evolution, but since Boyd does not like the concept of 'memes' he continues to speak of 'norms', sometimes in a rather ambiguous way.

The reviewer goes on to say:

That is one of the weaker points of Boyd's book: His poster examples from anthropological studies make perfect sense in a functionalist framework, as they are all about norms that have a strong (positive) effect on survival of HUMAN beings (useful food taboos, detoxifying material to obtain food, building shelters, etc..), but many, maybe even most of cultural variants (aka 'memes') do not have such effects. A lot of those entities culture is made of exist - and are reproduced - just because people find them interesting and memorable. That is exactly what is predicted by scholars who consider cultural variants aka 'memes' as a second replicator with its own independent 'fitness'.

I didn't read " A Different Kind of Animal" yet - but to be fair, this might be being a bit unfair to Rob. He's authored (or coauthored) many papers that use the "meme" terminology. He does understand that memes can be deleterious to their human hosts. There's a while chapter on this topic in "Not By Genes Alone". However constant use of "norm" as a substitute for "meme" does seem to be a kind of cancer in the field. Cultural evolution involves the study of memes, not just norms. Not all memes are norms.

Meme adjectives

A recent "Know Your Meme" editorial reminded me that I had never seen a list of meme categories or meme adjectives. I thought I would make a list of meme adjectives (loosely defined) - ordered by popularity:

• new memes (118,000,000 results)
• best memes (111,000,000 results)
• internet memes (46,400,000 results)
• funny memes (44,300,000 results)
• old memes (41,200,000 results)
• cat memes (40,500,000 results)
• girl memes (37,800,000 results)
• dog memes (32,100,000 results)
• great memes (23,200,000 results)
• boy memes (19,600,000 results)
• dank memes (13,400,000 results)
• viral memes (11,500,000 results)
• funniest memes (8,000,000 results)
• wholesome memes (7,720,000 results)
• fail memes (6,810,000 results)
• horror memes (6,870,000 results)
• fine memes (5,630,000 results)
• romantic memes (5,090,000 results)
• worst memes (3,540,000 results)
• greatest memes (3,140,000 results)
• asmr memes (2,080,000 results)
• creepy memes (1,940,000 results)
• moldy memes (1,590,000 results)
• deep-fried memes (343,000 results)

This is probably not an exhaustive Am I missing anything big? Can you improve on my methodology? Let me know in the comments. Thanks.

Dan Dennett: From Genes To Memes:The Evolution of Language and AI

One notable part is around 25 minutes in. Dennett complains about intelligent machines creating counterfeit people - and then proposes making counterfeit people illegal. No more Batman. No more Spiderman. No more Superman. It seems like an absurd proposal to me. What can Dennett be thinking about?

Memes start around 45 minutes in.