The following is the The 2021 global AI index reports that AI is being “industrialized” From Stanford University recommended by recordtrend.com. And this article belongs to the classification: artificial intelligence.
According to foreign media venture beat, China has surpassed the United States in the total number of artificial intelligence research papers, fewer and fewer artificial intelligence start-ups have received funds, and the U.S. Congress talks more about artificial intelligence than ever before. These are the three trends highlighted in Stanford University’s annual report, artificial intelligence index report 2021, released last week. The report attempts to document the progress of artificial intelligence and its impact on education, start-ups and government policies. The report details the performance progress of the main sub fields of artificial intelligence, such as deep learning, image recognition and object detection, and protein folding.
The AI index report is compiled by Stanford University’s human centered AI Institute and an 11 member Steering Committee with contributions from Harvard University, OECD, AI partnership and Sri international. The report uses data sets from a range of sources, such as arXiv’s AI research data, CrunchBase’s funding data, and surveys of groups such as black in AI and queer in AI. Jack Clark, former policy director of openai and head of an OECD Research Group on algorithm impact assessment, said the report also found a major trend in the industrialization of artificial intelligence.
“I think for me, the story is that artificial intelligence is being industrialized, and we don’t know how to evaluate its industrialization comprehensively, because we are a little short of a lot of data you expect. And I think that’s because AI has just gone from “not working” to “well deployed for business” faster than you think. It means Everyone is competing, including the research community, to keep up with the pace of business deployment, “he said.
Other main revelations in the report are as follows:
According to LinkedIn, Brazil, India, Canada, Singapore and South Africa have the highest AI recruitment levels from 2016 to 2020.
Global total investment, such as private investment and mergers and acquisitions, will grow by 40% in 2020. But for the third year in a row, AI start-ups have less and less capital flow to start-ups.
In 2019, in North America, about two out of every three graduates with a doctorate in artificial intelligence entered the industrial field, up from 44% in 2010.
Most of the AI PhD graduates come from outside the United States, and four out of every five graduates stay in the United States after graduation.
An analysis of 500000 blogs and 60000 English news shows that the AI ethics story is one of the most popular AI related news in 2020, including Google’s dismissal of timnit gebru and the ethical initiatives launched by the European Commission, the United Nations and the Vatican.
Attendance at major AI research conferences doubled in 2020 as most groups chose to hold virtual parties.
Women account for 18% of AI Ph.D. graduates, according to a 2020 survey by the society for computational research.
China has surpassed the United States in the total number of papers, but the United States has continued its leading position in AI research conferences for 20 years.
According to the total number of GitHub stars, tensorflow is the most popular AI software library, followed by keras and pytorch.
The number of AI related papers on arXiv has increased from about 5500 in 2015 to nearly 35000 in 2020.
The queer in AI 2020 member survey found that about half of the respondents have experienced harassment or discrimination and have encountered problems surrounding inclusiveness.
Academic researchers lead the world in the total number of published papers. But in the US, corporate research is second, while government research is second in Europe and China.
From 2004 to 2019, Carnegie Mellon University (16), Georgia Institute of Technology (14) and University of Washington (12) lost the most faculty to industry.
The section of the report devoted to the progress of technical challenges highlights the progress of computer vision systems and language models, as well as artificial intelligence for tasks such as drug discovery or effective chemistry and molecular synthesis.
AI index reports show the progress of AI systems that can be used for monitoring, such as the object detection system Yolo. Voxceleb has also made considerable progress, measuring the ability to recognize sounds from a dataset of 6000 people. AI index chart shows that the error rate of equality is about 8% in 2017, and it will drop to below 1% by 2020.
“This indicator tells us that the equal error rate of AI system has dropped from 8% to about 0.5%, which tells you that this capability will be quietly deployed on a global scale.” Clark said.
Alphafold’s ability to predict how proteins fold and gpt-3 are two of the most popular AI systems in 2020. Although AI index admits that gpt-3 has achieved little in learning, it cites a paper by timnit gebru and others, former co heads of ethical AI team, criticizing large-scale language models and their ability to continue bias. It also refers to a paper published last month by openai and Stanford University that argues that the social impact of large language models needs to be addressed before it’s too late. In an interview with VentureBeat in 2019, AI index founder Yoav Shoham questioned the value of judging language models based on the performance of limited tasks.
VentureBeat reports extensively on the two studies mentioned in the index. Other cited reports reported by VentureBeat include McKinsey’s the state of artificial intelligence report, which found that business leaders have made little progress in addressing the risks associated with deploying artificial intelligence. Another report warns of the dedemocration of AI in the age of deep learning, which co authors say may perpetuate inequality.
The AI index report includes calls for more benchmarking and testing in computer vision, ethics and NLP. As benchmarks like glue and superglue demonstrate, Clark said, “we’re running out of tests because we can build their speed.” Creating new benchmarks and tests is also an opportunity to enable metrics to reflect people’s values and to measure progress in addressing grand challenges, such as deforestation.
“I think one of the ways to get overall accountability in a space is to have the same test, you match everything to it, or the same set of tests. And before we have that, it’s going to be very vague to talk about bias and other moral issues in these systems, which I think will only hinder us as a community, but also make it easier for those who want to pretend they don’t exist or don’t mention them, “he said.
A few years ago, AI index expanded its scope to include arXiv monitor and other tools for searching preprint papers. AI index’s global vitality tool, which provides comparisons between national AI programs, is now applied to 26 countries in 23 categories.
This year, the report deleted data related to the progress of autopilot, and Clark said that due to lack of data, the report did not include information on fully autonomous weapons.
Stanford University: artificial intelligence and Economic Growth Research Report Stanford University: the number of Chinese AI papers published is the second in the world Automation, employment and productivity global AI Center: five factors driving the rise of AI in China? Beijing Zhiyuan Research Institute: Research Report on artificial intelligence in the next decade (download attached) Artificial Intelligence: opportunities and impacts of future decision making (download attached) US White House: artificial intelligence, automation and economy (download attached) from Bayesian theorem to probability distribution
If you want to get the full report, you can contact us by leaving us the comment. If you think the information here might be helpful to others, please actively share it. If you want others to see your attitude towards this report, please actively comment and discuss it. Please stay tuned to us, we will keep updating as much as possible to record future development trends.
RecordTrend.com is a website that focuses on future technologies, markets and user trends. We are responsible for collecting the latest research data, authority data, industry research and analysis reports. We are committed to becoming a data and report sharing platform for professionals and decision makers. We look forward to working with you to record the development trends of today’s economy, technology, industrial chain and business model.Welcome to follow, comment and bookmark us, and hope to share the future with you, and look forward to your success with our help.