Fluorescent protein, invented by AI

Nature Talk
1

The following is quoted from the beginning of an article, New glowing molecule, invented by AI, would have taken 500 million years to evolve in nature, scientists say:

An artificial intelligence model has created a new protein that researchers say would have taken 500 million years to evolve in nature — if nature were capable of producing such a thing.

As the article notes later on, the protein, named esmGFP, only exists as computer code at this time. An interesting and complex question is whether esmGFP actually could ever have evolved in nature. The article states that it is similar to those found in jellyfish and corals, and that it is:

… only 58% similar to the closest known fluorescent protein, which is a human-modified version of a protein found in bubble-tip sea anemones (Entacmaea quadricolor).

It also notes that esmGFP:

… would require a total of 96 different genetic mutations to evolve.

As we know, mutation and natural selection are both essential components of the process of evolution. For such a protein to actually evolve in nature, not only would the mutations need to occur, but it would also require that each state of the protein’s evolution along the way would need to survive natural selection sufficiently to lead to the next state. This would all need to occur in the context of actual organisms, ecological communities, and existing physical environments. Therefore it seems that any model that could provide a reasonable answer to the question of whether such a protein could actually evolve would need to be quite sophisticated, and therefore perhaps not possible, in practice.

The above complex issues are each elucidated by the article to some degree, but not in great depth.

Another interesting issue to consider is that AI could design many other such complex synthetic biological substances that have never evolved, and whether there is the possibility that some, if actually created and exposed to living environments, could be taken up by living organisms, and become destructively invasive.

So, would anyone like to offer any interesting thoughts about any of the above?

2

I am not a biologist, but is this not basically true of most introduced organisms? Things that are introduced where they do not belong can wreak havoc.

It appears the flex point of this is “if actually created”, and I cannot know whether or not that is possible, and then the question is if created, whether or not it would actually sustain and if so if it would do so beyond a sterile laboratory setting or fall to pieces like so many other things that seem to work on paper or in design.

To be honest, I worry far more about things that already exist and can do so, such as historical pathogens.

1 Like
3

Yes, I agree that for the reasons you cited, introduction of existing organisms to new environments is currently a much greater problem. Odds are that any single substance designed by a computer and subsequently synthesized would probably not replicate in the natural environment, and therefore would not present a problem. On that basis, it is highly unlikely that esmGFP is a worry.

But a concern is that computing technology is developing rapidly. If a system were to be developed that could rapidly design and crank out actual samples of synthetic proteins and other biological substances by the thousands, a single accidental release of a cache of these could include one or more substances that pose a genuine hazard.

1 Like
4

I believe if the organism really needed that protein it would not take 500 million years to evolve. We are being lulled by the term artificial intelligence. There is nothing artificial about it. It is real and it is evolving very quickly. I see it as another form of Non Human Intelligence which already exists in all living organisms that share the earth with us.

I find it difficult to believe random mutations can produce the diversity and complexity of life. It seems mathematically impossible given the slow rate of mutations. The following are some of the questions I wonder about when I’m doing my nature walks. Has some form of intelligence evolved in the DNA molecule to produce changes in organisms? It seems to me any early organism with even a slight capability to modify its form would have a tremendous evolutionary advantage. Do organisms evolve individually or jointly? If jointly as a system, is there any coordination (communication) of intent or direction? When I stare at a mantis or a jumping spider I wonder what is it thinking. It’s looking at my eyes so it knows what eyes are. How does it know that? Which brings up another thought. The evolution of a new physical characteristic also needs to evolve the knowledge and behaviour needed to use the new feature. Do mature adult organisms pass on learnings from their life experience to offspring. In other words, are important learnings and abilities saved in the DNA and passed on to the progeny and is this what we call instinct?

A bit of a ramble and I don’t expect answers but would appreciate any links for further readings on the evolution of intelligence. But with respect nothing on Creation.

3 Likes
5

We could speculate about what underlies the fundamental laws of the Universe. However, is there any experimentation possible that can verify any such hypotheses? That’s not meant as a rhetorical question, but is rather an expression of my inability to come up with any suggestions for a verification process.

1 Like
6

Yes it’s all speculation but a good mental exercise. The fossil record helps with the physical aspects of evolution but not the early evolution of intelligence. It also doesn’t help that our intelligence (in its broadest definition) is very different than the intelligence in most of the life forms out there.

1 Like
7

Studying that topic would generally involve considering the evolution of a nervous system and brains. Suggestion of intelligence in a DNA molecule would be greeted with some skepticism.

Indeed, it is a complex protein that has not been produced as an actual material. It may be that a method for producing it will never be developed. An automated system for designing and subsequently producing such complex materials would require a sophisticated system of containers, heaters, distillers, condensers, mixers and other chemistry equipment, all coordinated by a computer system, so it may be some time before such a design and production system is possible. If it ever becomes feasible, its potential consequences should be carefully considered, just as how we currently need to think about how to manage AI.

Another protective barrier may be that many of these substances, if released, might not actually replicate.

8

And yet a DNA molecule, like a line of computer code, is encoded information. That makes it fundamentally different from, say, a crystal lattice.

2 Likes
9

It’s hard to find material that skirts the line between these thoughts but this article I came across recently might qualify, particularly the supplement at the bottom: https://www.thenewatlantis.com/publications/evolution-and-the-illusion-of-randomness
I’m not familiar enough with molecular biology to evaluate the quality of the arguments there though.

1 Like
10

Thanks for the article. The author does a great job of explaining the problem and the references at the end provide a good list for further reading.

2 Likes
11

Thanks for the link to the article. It presents much to think about, and I’m reading it slowly and carefully, with the intention to offer some thoughts about it here in a few days. Soon after beginning the article, it sparked a memory of my having read material by Daniel Dennett several years ago. Then, sure enough, he gets discussed later on in the article. I’ve never quite understood his stance regarding our own consciousness, and it seems to me that he claims that it is just an illusion that it even exists. That doesn’t make any sense to me. Perhaps I’m mistaken about his stance on consciousness, and if so, someone here can correct me.

1 Like
12

I don’t know his position, but I gather the standard physicalist explanation of consciousness would be that it’s an emergent phenomenon (as with life itself). I feel like it wouldn’t be accurate to say that emergent phenomena are illusions though. That seems like saying “things that are too complicated for us to understand are actually just simpler things that seem complicated.” To me it seems obvious that consciousness is real, and immensely complicated. Just because something can’t be probed or measured doesn’t mean it’s nonexistent.

2 Likes
13

It is like opening the pandora box, or like Stephen hawking said, AI may doom mankind. I can’t count. Assuming the present mankind did not exist. Another form of mankind come into existence at a point in time. It will be very random when he/she could invent an advanced machine which run through some permultations using the machine to pick out a novel protein. I doubt the 500 million years figure, because it is like winning lottery.

1 Like
14

The theory of evolution may have answers to your questions. The beginning of life on earth will always be a mystery.
The rates of mutation is not constant. If an organism is living in a radioactive area, it is subjected to higher radiation levels. The sands, rocks, potassium all said to have radiation. Sunlight has radiation. Various environmental circumstances shape an organism. A creature on earth does not evolve alone. It is predated upon, or it eats another organism. Plants grow in a certain way in response to animals eating its leaves. It is amazing some bush turkey chicks come out of its shell and it is on its own at that stage. While domestic chicken takes care of its chick for a while. I have seen the chicken feeding its chicks. Instincts is built-in maybe in the DNA.

1 Like
15

Yes. It seems a contradiction to assert that it is an illusion, since it would leave one at a loss to explain who it is that is having that illusion. To be accurate, it does seem that the article takes issue with at least some of what Dennett claims. Within this article, that contention is not particularly about our consciousness, but rather about some of what Dennett claims regarding evolution.

There is much subtlety in the article, therefore it deserves a careful reading. Consider this:

Or as Douglas Futuyma, distinguished professor of ecology and evolution at the State University of New York at Stony Brook, once put it: “Mutation is random in [the sense] that the chance that a specific mutation will occur is not affected by how useful that mutation would be.”

Note the “in [the sense]” qualifier that would still allow for frequencies of mutation to vary among different types of organisms. There may be some advantage to producing mutations at a rate that simultaneously maintains some integrity of fitness, while also allowing a steady rate of emergence of new forms, some of which could turn out to be advantageous. Sometimes, at least in educational settings, mutations are characterized as “mistakes” in making copies of chromosomes, however, such a characterization presumes that the purpose of the process is precise duplication, which is an overly bold claim. There is some subtlety there in the precise meaning of “mistake”.

More to say later …

2 Likes
16

The article linked from post 9 of this discussion sets the stage for what follows with its title “Evolution and the Illusion of Randomness” and subtitle “Survival, fitness, and the purposiveness of organisms”. Essentially, the article seeks to oppose the notion of evolution as a process lacking in purpose. It also opposes the notion that mutation is essentially random. In addition, it notes the difficulty defining fitness, and characterizes the prevailing concept of it as tautological.

That the article seeks purpose as an important component of evolution is underscored by this paragraph within the ‘Couldn’t You Be More Explicit Here?’ section of the article:

In any case, it is startling to realize that the entire brief for demoting human beings, and organisms in general, to meaningless scraps of molecular machinery — a demotion that fuels the long-running science-religion wars and that, as “shocking” revelation, supposedly stands on a par with Copernicus’s heliocentric proposal — rests on the vague conjunction of two scarcely creditable concepts: the randomness of mutations and the fitness of organisms. And, strangely, this shocking revelation has been sold to us in the context of a descriptive biological literature that, from the molecular level on up, remains almost nothing but a documentation of the meaningfully organized, goal-directed stories of living creatures.

It is already understood among biologists that fitness is challenging to define precisely. It is also recognized that not all mutations have an equal likelihood of occurring.

As humans, we perform many activities with purpose, as do other animals, therefore purposiveness is a part of existence. However, the article also seems to seek purposiveness in the process of evolution itself. However, if doing so, we need to be careful not to form conclusions about where purposiveness occurs that might lead us to make claims about evolution that run counter to evidence.

As for the claim by scientists described by the article linked from the original post that the modeled protein would have taken 500 million years to evolve in nature, it does depend on assumptions about the relative likelihoods that the modeled mutations would occur. In that regard, and the lack of taking natural selection into account, it is difficult to demonstrate that the molecule ever could evolve. The article, though, does admit the possibility that it might not actually ever be possible for that to occur.

Both articles are interesting, but also need to be understood in the context of the limitations of the concepts they present.

3 Likes