【推荐1】The average person must think the field of artificial intelligence or AI is making great progress. The boss of a top tech company thinks that in search of creating AI that has the flexibility of the human brain, “the game is over”. Don’t be fooled. We should see it in a reasonable way. Machines may someday be as smart as people and perhaps even smarter, but the game is far from over. There is still a large amount of work to be done in making AI machines that can truly understand and reason about the world around them. What we need right now is more basic research to speed it up.

AI is making progress. Artificial images look more and more realistic, and speech recognition can often work in noisy environments. But we are still likely decades away from human-level AI that can understand the true meanings of articles and videos or deal with unexpected difficulties. The field is stuck on the same challenges that scientists have been pointing out for years: getting AI to be reliable and getting it to flexibly deal with unusual situations. Unluckily, the fact is that these systems still fail to work reliably and struggle with new situations.

Although deep learning has advanced the ability of AI machines to recognize patterns in data, it has three main disadvantages. The patterns that it learns are superficial. The results it creates are hard to explain. And the results are difficult to use in the other processes, such as memory and reasoning. As Harvard University computer scientist Leslie Valiant noted, “The central challenge in the future is to unify the formulations (统一公式) of learning and reasoning.” You can’t deal with a person carrying a stop sign if you don’t really understand what a stop sign even is.

It’s time for AI researchers to get out of turning to the media straightly for help, and ask important questions about how to build systems that can learn and reason at the same time.

【推荐2】An artificial intelligence (AI) trained on personal data covering the entire population of Denmark can predict people’s chances of dying more accurately than any existing model, even those used in the insurance industry. The researchers behind the technology say it could also have a positive impact on early prediction of social and health problems.

Sune Lehmann Jorgensen at the Technical University of Denmark and his colleagues used the data covering various aspects of 6 million people from 2008 to 2020 from Denmark. They changed the data into words that could be used to train a large language model, the same technology that powers other AI apps. These models work by looking at a series of words and determining which word is statistically most likely to come next, based on vast amounts of examples. In a similar way, the researchers’ Life2vec model can look at a series of life events that form a person’s history and determine what is most likely to happen next.

In experiments, Life2vec was trained on all but the last four years of the data, which was held back for testing. The researchers took data on a group of people aged from 35 to 65, half of whom died between 2016 and 2020, and asked Life2vec to predict who lived and who died. It was 11 percent more accurate than any existing AI model used to price life insurance policies in the finance industry.

Jorgensen believes that the model has consumed enough data that it is likely to be able to shed light on a wide range of health and social topics. This means it could be used to predict health issues and catch them early, or by governments to reduce inequality. But he stresses that it could also be used by companies in a harmful way. “Clearly, our model should not be used by an insurance company, because the whole idea of insurance is that, by sharing the lack of knowledge of somebody who is going to be the unlucky person struck by some incidents or deaths, we can kind of share this burden,” says Jorgensen.