The evolution of ideas – how artificial intelligence can help us
People want to know how stories continue. This is true of streaming series, and it is certainly true of the biggest story of them all: the evolution of life on Earth. Species change, and anyone or anything that once ruled the Earth can also perish – or embark on the journey from dinosaur to chicken. So what will become of humankind? What will it have to make way for?
Artificial intelligence (AI) comes into play in many of the visions of how evolution will continue. In the gloomier versions: in the evolutionary race, humankind loses out to machines which become ever faster and have long since spun out of control. Humankind perishes … but can at least pride itself on having brought the new rulers into the world. A somewhat more positive forecast is provided by AI pioneer Jürgen Schmidhuber. According to this forecast, the super-intelligent beings leave Earth behind them and set off into space because, quite simply, that is where more abundant sources of energy are to be found.
But maybe humans won’t need to yield to machines at all, but can instead join forces with them? If we go along with the thinking of Paul Rainey, who works at the Max Planck Institute for Evolutionary Biology in Plön, all we would have to do is make humans equip themselves with AI devices which accompany them their whole life long, and then copy the content of these devices onto their children’s own devices. This would get around the difficulty of AI systems being able to improve themselves but not to procreate. As independent cells once became organelles of larger cells, AI devices and humans could become a larger, more intelligent and – in the evolutionary sense – a better adapted unit of evolution.
More radical visions see humans on a path towards becoming cyborgs, pursuing the same goal through an increasing number of technical implants – namely, becoming ever more intelligent and living ever longer. Natural selection would have been largely eliminated and, instead, humans would have taken their further development into their own hands. An extreme form of this fantasy would provide for human consciousness being “uploaded” into a machine and thus becoming immortal. In this way, evolution – or at least the evolution of humans – would have arrived at its final point.
Other people in turn are thinking ahead in developing AI processes which are already being used anyway in the field of (evolutionary) biological research. With the aid of learning algorithms, evolutionary processes can be traced much more accurately on the basis of large quantities of data, evolutionary lineages can be refined, and the importance of certain genes for evolutionary developments can be established.
Taking this approach, it is possible not only to look back into the past – it can also be used to forecast how evolution might continue, simulate the development of populations under different conditions, or assess any possible evolutionary consequences of genetic manipulations.
Scenarios relating to possible applications range from attempts to equip plants better to cope with climate change to the idea of optimising humans not with electronic components but by means of newly developed proteins or entire genes. Taken to extremes – and this has already been discussed – humankind could split up into different species: people who can afford it have themselves genetically improved, while the others will have to be content with their natural genetic makeup.
The process of evolution is complex, and algorithms trained with large quantities of data promise to make this complexity comprehensible and manageable. While the transfer of consciousness into a machine is just as fantastical as any autonomous development of robots, there will be experiments to change humans by means of biological and technical methods – and, it is to be hoped, with due prudence. Just how far this will impact on humans’ evolution is unforeseeable – however curious we may be on how the evolutionary process continues.
Perhaps we should let the big visions take a back seat and get AI help us to think more out of the box – testing architectures, combinations of materials or experimental structures which, at first sight, make no sense, but which may in fact be just right. This way, AI may not provide us with a leap forward in human evolution, nor with supersmart successors, but it may present us with some good ideas. We could do with such ideas to solve problems which are currently really important for the survival of our species – from climate change to peaceful co-existence.
This article is from the University newspaper wissen|leben No. 5, 17 July 2024.