Artificial intelligence to find new molecules
“It is estimated that there are 1,060 molecules that can give potential drugs, more than all the atoms in the solar system, announces MIT.
This offers virtually unlimited chemical possibilities.” In order to explore this ocean of molecules, the researchers are relying on artificial intelligence (AI), which will sift through the molecules and relate them to potential targets, such as a protein or a cell receptor.
In July 2019, an Australian team designed a flu vaccine with an algorithm-designed adjuvant. And, in February 2020, the company Insilico Medicine managed to develop a drug against fibrosis in just 46 days thanks to its AI, where the development of a conventional drug takes up to 10 years.
After 120 satellites launched last year, SpaceX plans to deploy up to 42,000 to create an Internet connection anywhere on the planet.
The company is not alone in the niche: the One Web constellation will include 600 satellites by 2022 and Amazon has announced that it wants to launch 3,236 satellites in low orbit to cover white areas. All this thanks to the low cost of launching these nanosatellites which weigh barely a few kilos.
The deployment in space of so many objects, however, poses problems in terms of interference with other satellite services such as weather, increase the risk of collision and disrupt astronomical observation.
It has been more than 50 years since we announced the arrival of the quantum computer. These machines, where the bits are replaced by qubits with superimposable states, are in theory much faster and more efficient, but until now suffered from decoherence problems.
In October 2019, however, Google announced that it had achieved quantum supremacy (superiority of a quantum computer over a classical computer on a particular task) by performing a calculation in three minutes that would require around 10,000 years on a classical supercomputer.
A figure disputed by IBM, which estimates that a conventional computer program could have solved it in just 2.5 days. Nevertheless, we are witnessing the take-off of the quantum computer, with major players such as Microsoft, D-Wave, Atos and the CEA all jumping into the race.
The computing power needed to train artificial intelligence algorithms doubles every 3.4 months, according to a study by Open AI. In addition, the supercomputers dedicated to these programs require a gigantic consumption of energy.
The digital giants are working today to miniaturize the technology to make it accessible to the general public. Google Assistant and Siri thus integrate voice recognition systems on a smartphone chip.
AI is also making its way into digital cameras, capable of automatically retouching a photo by removing an annoying detail or improving the contrast, for example. “Localized [rather than cloud-based] AI is better for privacy, since your data no longer needs to leave your device,” MIT said.
How to massively collect personal data without infringing on privacy? This is the challenge for the US Census Bureau, which will have to secure 330 million profiles for its 2020 census, so that it is impossible to identify each individual.
For this, it will inject “noise” into the database, in order to complicate a possible de-anonymization. This method called “differential privacy” is already used by Facebook and Apple to aggregate data without reaching the exact identity of individuals.
This scrambling system is fundamental for many sectors using sensitive data, such as medical research.
Taking climate change into account
Until now, even the experts were reluctant to attribute any climatic catastrophe to global warming. This is starting to change, and we are now able to model the exact role of climate change.
In particular, this should enable insurers to anticipate and distribute the costs of floods, storms or droughts. The Central Reinsurance Fund (CCR) in partnership with USA Weather report has calculated that the cost of claims should double by 2050, with climate change accounting for 20% of this increase.
Thanks to detailed satellite data, it is also able to model a tsunami or predict how much the water will rise during a flood.
The new nuclear
Reviled by environmentalists, nuclear energy is nevertheless an energy of the future, believes Bill Gates, who also supports several start-ups in this sector.
But not “old school” nuclear power like that of the EPRs: the billionaire is betting on small modular reactors (SMR), presenting much less financial and ecological risks, like those of NuScale. Research on nuclear fusion could also soon be successful, providing almost unlimited zero-carbon energy.
If current robots are very good at performing repetitive tasks on an assembly line, they are not very agile at manipulating objects individually according to their shape or context as humans do.
By combining visual recognition, deep learning algorithms and robotics, robots will be able to do as well as they do tomorrow, hopes Bill Gates, who cites the example of Dactyl, a robotic “hand” developed by the Open AI association.
The personalized cancer vaccine
With nearly 10 million deaths each year, cancer is the second leading cause of death worldwide after cardiovascular disease, according to the World Health Organization (WHO).
Among the hundreds of new therapeutic approaches, that of the German startup BioNTech caught the attention of Bill Gates. This start-up detects mutations in cancer cells and uses patients’ own immune systems to develop vaccines targeting these mutations.
The endoscope pill
Have you ever heard of environmental enteropathy (EE)? This disease, which results in inflammation of the intestine and leads to malnutrition, causes retarded growth and psychomotor development.
Detectable only by endoscopy, it could however be better diagnosed thanks to an electronic pill to be swallowed which will inspect the intestine and perform biopsies.
Developed by Guillermo Tearney, biologist and engineer at Massachusetts General Hospital (MGH) in Boston, it can also be used to treat other pathologies, such as colon cancer.
Meat without meat
In 2050, the planet will have 9.8 billion inhabitants who will consume 70% more meat than in 2005. However, the production of one kilogram of beef requires 70 times more surface area, eight times more water and emits nine times more earth greenhouse gases than the production of one kilogram of soybeans.
Cultivated in vitro with stem cells or based on plants, “fake” meat is therefore the subject of big ambitions on the part of startups such as Beyond Meat or Impossible Foods, in which Bill Gates has also invested.
Toilets in developing countries are one of Bill Gates’ great hobbies. Some 2.3 billion people today still do not have access to clean sanitary facilities, he constantly reminds us, which encourages people to defecate in the open and promotes the spread of disease.
In 2011, the philanthropist launched the “Toilet Challenge” calling on innovators around the world to create low-cost, autonomous toilets. Dozens of prototypes have already been presented.
ECG connected watches
Last May, Apple proudly announced that its connected watch had saved the lives of two people by detecting an abnormally high heart rate. The latest version of the Apple Watch is indeed equipped with an electrocardiogram (ECG) function analyzing cardiac impulses.
However, current devices with a single sensor cannot detect a stroke in real time. This could soon change: the American manufacturer AliveCor presented at the last Congress of the American Association of Cardiology a device with two sensors supposed to detect certain types of attack.
To prevent global warming, it would be necessary to remove 1,000 billion tons of carbon during the 21st century, according to UN experts. As recently as February 26, Australian researchers presented a method to capture and convert atmospheric CO2 into storable solid carbon.
Bill Gates cites the startups Carbon Engineering (in which he invested) and Clime works, which transform atmospheric CO2 into biofuel or industrial gas.
“It’s still expensive and difficult, but given our inability to sufficiently reduce our emissions, there is no other option”, explains Bill Gates, fatalist.
Predict premature births
Fifteen million babies are born prematurely every year, one million of whom will die from complications, according to a report from some 50 organizations and universities.
“Tomorrow, a simple blood test will show whether a woman is at risk of premature delivery,” says Bill Gates.
Stephen Quake, a researcher at Stanford University, has developed a test that detects the expression of certain circulating RNA genes (outside cells) associated with a risk of premature birth. Mothers at risk could thus be identified and cared for.
Smart voice assistants
Amazon Alexa, Siri, Google Home… Connected speakers have been widely adopted in the United States, but their capacities remain quite limited for the moment.
The deeper understanding of human language combined with the improvement of synthetic voices will soon make it possible to have a real conversation with voice assistants, assures Bill Gates, who cites in particular the example of Google Duplex, capable of phoning to make an appointment. you in your place.