Originally published on: https://zhcn.kevinhu.me/2021-hai-ai-report/

Report address: https://aiindex.stanford.edu/report/

Stanford's Human-Centered Artificial Intelligence, co-founded by John Etchemendy and well-known Chinese professor Feifei Li in 2019, aims to "point the direction for the future development of this complex new field of artificial intelligence". Its work includes research, education, policy-making that promotes AI learning, studying how AI can benefit people, advocating for everything to be "human-centred," understanding and predicting how AI will affect human society, and promoting development that empowers human beings artificial intelligence, rather than replacing humans.

Reference: https://aiindex.stanford.edu/about/

The direction of HAI research focuses on the application of artificial intelligence in society, and pays more attention to social benefits and human impact. It pays more attention to the topics of artificial intelligence ethics, such as deepfake technology (deepfake), and gender and racial discrimination in artificial intelligence results. sensitive.

As a new technology, artificial intelligence has begun to have a significant impact on human society. We can't fully foresee the details of how it will change our society, but one thing is for sure: it will certainly make a significant difference to our society, but we cannot sit idly by and let it bring harmful changes to our society. We must always be on the lookout for new negative effects it may bring.

If you are an artificial intelligence practitioner, a technologist interested in the technology, a sociological researcher, or someone interested in the future impact of artificial intelligence on society, HAI's blogs and reports may be able to help you.

HAI produces an annual report on the development of artificial intelligence. The 2021 report provides observations and summaries of developments over the past year. Overall, AI is booming and continues to grow exponentially in metrics such as the number of papers and conferences. The number of students and researchers in higher education institutions attracted by AI majors also continues to grow. There have been breakthroughs in several important research fields: such as openAI's GPT-3 natural language model to generate highly credible text content, the Google Brain research project has enabled humans to predict protein molecular structures with high precision for the first time through machine learning, etc. .

The full report contains more details, from which HAI summarizes nine megatrends to watch:

1. Artificial intelligence is heavily invested in applied drug design and discovery

Affected by the new crown epidemic, artificial intelligence has been widely used in the fields of drug design and biomedical research. "Medicine, cancer, molecular drug research" and other fields have received the highest investment in all artificial intelligence application fields in 2020, with more than 13.8 billion US dollars, more than 4.5 times more than in 2019.

2. Employment continues to tilt towards the AI industry

In 2019, 65% of North American PhD graduates in artificial intelligence went on to work in industry, up from 44.4% in 2010. This may mean that artificial intelligence has moved from a more academic research direction to an industrial application direction.

Third, artificial intelligence can generate a variety of content

Artificial intelligence is already capable of generating vast amounts of content: text, audio, images, and more. It is already difficult for humans to distinguish whether these generated contents are generated or not under certain conditions.

Fourth, artificial intelligence still has the problem of diversity

Minorities are still underrepresented in AI. In 2019, 45% of Ph.D. graduates were white, compared to 2.4% black and 3.2% Hispanic.

5. The number of Chinese papers surpasses that of the United States

This is good news for China: China has surpassed the United States not only in the number of AI research papers, but also in total citations.

Sixth, more than half of the researchers with doctoral degrees in artificial intelligence in the United States are from abroad

More people come to the United States from overseas to study artificial intelligence, and the United States is still attracting talents effectively. In 2019, 64.3% of PhD students in new AI directions in North America were from overseas. A large number of them choose to continue to work in the United States after graduation.

Seventh, the monitoring technology using artificial intelligence is constantly improving and it is more widely used

Surveillance technology using artificial intelligence is still being enhanced, such as image classification technology, facial recognition technology, video analysis technology, and speech recognition technology, etc. have greatly improved.

8. There is still no good benchmark and consensus on AI ethics

Although many organizations have launched multiple quantitative or non-quantitative tests in the field of AI ethics, there is still no consensus standard for the entire field. That said, AI ethics still doesn’t have a good yardstick to test how much social discussion of AI ethics affects real technology development. But it can be observed that academia and society generally pay more attention to ethical issues than industrial fields.

9. Artificial intelligence has caught the attention of the U.S. Congress

This session of the US Congress has also paid more and more attention to artificial intelligence. There are three times as many mentions of artificial intelligence in Congress minutes than in the previous Congress.

summary

According to the report, the general development trend of artificial intelligence in the past few years is unchanged, and it may be maintained for many years to come. It will:

• continue to attract more attention, research, investment, and learning;
• Have a stronger ability to make more effective contributions to scientific and technological research and social applications;
• It will further challenge our understanding of the ethics of technology and our response to and decision-making about the negative effects of new technologies.

Artificial intelligence may be one of the most important technologies of this era. Like all important technologies of the past, it will bring unprecedented changes and transformations to our society and the world. It will give us new perspectives and tools to observe, study, and change the world, affecting all aspects of human research, medical care, transportation, education, entertainment, etc., and become part of our daily lives. We have to wait and see how the world and artificial intelligence will get along in the future.