Artificial intelligence (AI)-based medicine is no longer just a vision of the future. After U.S. Secretary of Health Robert F. Kennedy announced investments in medical database programming over the summer, there were hints that AI will one day diagnose patients and prescribe drugs on its own.

In early November, Meta chief Mark Zuckerberg and his wife Priscilla Chan announced the expansion of Biohub, a medical initiative dedicated to prevention and the "cure for all diseases". As a pediatrician, Chan regularly cites her profession as a source of motivation.

Artificial intelligence and real-world diagnostics

Biohub is a non-profit research organization that aims to eliminate common diseases by the end of this century. It is subordinate to the Chan Zuckerberg Initiative, which is the main beneficiary of the couple's profits from Facebook's messaging platform.

It collaborates with universities in the Californian cities of Berkeley, San Francisco and Stanford, using a system for visualising human cells in the universities' laboratories. The latter serves as an "experimental dummy" to allow scientists to observe the onset and development of diseases without the use of living cells.

In a video from early November, Zuckerberg boasted that scientists are working with cutting-edge artificial intelligence and "the largest datasets of human cells" as part of the expanded initiative. "The next decade is going to be really exciting, and we think some of humanity's biggest dreams will finally become achievable," the Facebook founder added.

Back in late July, it was Zuckerberg who declared on his blog that Meta was starting to see "glimpses of superintelligence" that was "improving itself". If such an autonomous AI model enters into cooperation with the human genome database, it can start suggesting modifications at any time - which will happen sooner or later.

Secretary Kennedy is controversial in the eyes of the US and world media, mainly because of his sceptical attitude towards vaccines, which, unlike his predecessors, he does not consider a "panacea". However, at the end of July, he signed an agreement with companies such as Google, Anthropic (author of the Claude model), OpenAI (developer of ChatGPT), Amazon and Apple, which is to create an all-American database of health data.

This is to fall under the Center for Medicare Services (CMS) and, according to both the department and the White House, is to be "powered" by artificial intelligence. This means that one of the language models is supposed to facilitate searches in it - similar to the way the Gemini model calculates answers when words are typed into the Google search engine.

However, researchers whose focus is similar to Biohub envisage a much broader scope for such an AI model. ChatGPT and similar models could, at some point, self-diagnose a patient and issue a prescription for a drug.

Diagnosis of rare genetic diseases is still in its infancy, with individual language models not equally accurate. However, their access to Biohub's database could train them to a completely different level than they have been.

So will AI models start to suggest genetic modifications to prescription drugs? The CRISPR-Cas9 gene-editing system has been patented since 2013, and the entire human genome has been sequenced since the days of President Bill Clinton.

The first genetically modified baby was born in 2019, and it was born in China. Doctors have also used technology like CRISPR to recode genes that could cause unspecified civilization diseases, so the philanthropic motivation cannot be denied.

"We are creating cutting-edge artificial intelligence for biology, trained on our vast and unique biological datasets, to better predict how human cells behave andhow they might change," Biohub announced bluntly, adding that scientists will use Meta's AI models for experiments and discoveries "in the field of human health and disease."

"The results will be able to be fed back into the AI models, improving their predictive ability - and ultimately allowing diseases to be treated." But while ethics in AI focuses on how to prevent models from "lying," here a very different ethical threat opens up.

Just as erythropoietin-based anaemia drugs are being used to boost red blood cell production in doping, so AI-generated model cases of genetic defects may one day begin to be used to "enhance" normal human abilities.

What this will mean for human beings and human nature as a whole, everyone will have to answer for themselves.

Space as the final frontier

Another technological discovery this month, while not threatening the very definition of humanity, has the potential to threaten our environment and is located in China.

The team at DFH Satellite, led by Su Zhenghua, has produced a device that generates electromagnetic pulses - those that accelerate ejected particles - with the required precision, without losing power.

"Existing pulsed power sources typically have an output power of less than one megawatt and synchronization accuracy of worse than one millisecond, with limitations in current control accuracy and conversion efficiency," Su and his colleagues wrote in a peer-reviewed study in the journal Advanced Small Satellite Technology.

At issue is a satellite system that generates high-energy particle beams similar to the Death Star from the iconic Star Wars franchise. The device, which emits photons, electrons or other subatomic particles at high frequencies, is designed to work just like the laser from the famous sci-fi franchise.

An adversary's space device - whether it be a rocket, satellite or anything else - could become the target of kinetic or thermal damage. However, high energy and high precision are mutually exclusive, making this concept science-fictional.

DFH's device achieves 2.6 megawatts of pulsed power, while maintaining synchronisation at 0.63 microseconds. Both the study and the successful ground tests were highlighted by the prominent South China Morning Post (SCMP).

The common reader in the West will automatically think of the Death Star, but the Morning Post noted that the device also has potential non-military applications - particularly in lidar technology [the creation of 3D relief maps via laser, editor's note]. At the same time, however, it acknowledged a new series of space races.

"China is plunging into the development of high-performance space energy systems at a time of growing strategic competition in orbit. As the U.S. expands its Starlink constellation and planned Starshield - a network of thousands of small, durable, dual-use satellites - traditional weapons such as missiles are becoming less practical for space defense," the SCMP wrote bluntly.

The newspaper also floridly described the advantages of laser weapons, especially in terms of cost, since "shooting down cheap, fast-moving satellites with expensive missiles is inefficient and unsustainable."

"Directed-energy weapons such as lasers and particle beams could disable or damage multiple targets at the speed of light, using only the electricity produced by solar panels. The cost per shot is almost zero," the Morning Post added.

But there is a civilian side to the space race. Nvidia has entered this chapter, and although it was labelled an investment bubble by economists earlier this year, it has been involved in the development and successful use of an AI space signal finder.

UC Berkeley astronomers offered their Allen radio telescope to the Breakthrough Listen initiative, which deployed an AI model to scan and decode Fast Radio Bursts (FRBs).

Berkeley shares management of the Allen Radio Interferometer Array - an array known as the Hector Telescope - with the Search for Extraterrestrial Intelligence Institute (SETI). The latter is looking into probable radio signals from distant parts of the universe that could be the result of intelligent life.

The deployment of a model based on Nvidia's Holoscan platform has seen up to a 600-fold increase in computing speed. Unlike the method previously used, called dedispersion, it scans huge amounts of data in real time and eliminates up to 90 percent of false positives.

It is a process that involves decoding radio signals using a phase shift against the background, which in practice slows down research. However, the modified Holoscan has scanned thousands of FRB signals up to 160 times faster than real-time.

In tests, the AI successfully detected huge pulses from the Crab pulsar, processing data streams at up to 86 gigabits per second.

SETI published a press release on November 5, the closest interstellar object to the Sun at the time of the 3I/ATLAS event. Its suspicious trajectory and behaviour during its journey have been in the media several times, and it will be beyond Mars' orbit by November 25. The Institute has not picked up any signal from this comet or asteroid.