Inside | Real news, curated by real humans
Inside AI

Inside AI (Dec 30th, 2019)

Let's take a look back at the top 25 stories of AI in 2019. Inside returns to its regular publishing schedule on Jan. 2. Happy new year.

– Beth

1. Scientists from the University of California, Berkeley, and Google Brain developed an AI system that taught a robot to walk. The team used a "maximum entropy" reinforcement learning framework to learn locomotion skills without simulated training. - VENTURE BEAT

2. China and the U.S. are leading the global AI race, according to a study by the U.N. World Intellectual Property Organization (WIPO). The report found that U.S. tech company IBM had, by a wide margin, the most AI patents with 8,920; Microsoft was second with 5,930. - REUTERS

3. Summit, the world's most powerful supercomputer, broke a record for the fastest-running machine learning experiment. The experiment used algorithms to detect extreme weather patterns such as hurricanes out of a huge dataset of climate simulations. - WIRED

4. Trump signed an executive order in February launching the "American AI Initiative." The national AI strategy includes five key goals: redirect funding, create resources for researchers, establish standards, retrain workers, and engage internationally. - TECHNOLOGY REVIEW

Articles: TECHCRUNCHPCMAGVENTURE BEATWIRED

More: Shortly after Trump signed the executive order, the Pentagon unveiled its own AI strategy to help the country compete against China and Russia. The strategy centers on the newly formed Joint Artificial Intelligence Center, led by Lt. Gen. Jack Shanahan. Intel also released its own response to the American AI Initiative in a white paper back in March.

5. OpenAI developed its unsupervised language-learning model - GPT2 - that can write news and answer reading comprehension problems. The system learned from reading 8 million internet articles and set a performance record on Winograd schemas. It can also generate text. - VOX

6. China's Xinhua News Agency unveiled a female AI news anchor. Xin Xiaomeng launched in March after Xinhua debuted its male AI news anchor Qiu Hao. - QUARTZ

7. Chinese chip maker Horizon Robotics raised $600 million in a Series B funding round. The funding was led by South Korean chipmaker SK Hynix and several auto manufacturer-backed VC funds. - REUTERS

8. In March, OpenAI announced that it's restructuring into a for-profit institution. The nonprofit AI organization began calling itself a "capped-profit" company that can issue stock to its employees and generate returns. Most of the money earned is returned to the nonprofit entity. - BUSINESS INSIDER

9. MIT researchers developed an algorithm that can build a neural network much faster — up to 200 times faster — than conventional methods. The Neural Architecture Search (NAS) algorithm can learn specialized convolutional neural networks for specific hardware platforms in just 200 GPU hours. - ENGADGET

10. A group of AI researchers, including a Turing Award winner, asked Amazon to stop selling its biased facial-recognition technology to law enforcement agencies. They signed an open letter saying that the company's algorithms are flawed and biased against darker-skinned and female faces. - NY TIMES

11. To train Alexa to understand human speech better, Amazon instructs thousands of employees to listen to and annotate customers' Echo voice recordings, according to an April report from Bloomberg. The Amazon team listens to thousands of recordings per day. Amazon said it annotates "an extremely small sample of Alexa voice recordings" and users can opt out in Alexa's privacy settings. - BLOOMBERG

12. San Francisco became the first U.S. city to ban the use of facial recognition technology used by law enforcement agencies. - CNET

13. American companies are looking abroad to find enough skilled workers in AI. Schools, universities, and training programs haven't updated their offerings enough to keep up with market demand. - WSJ

14. AI researchers began more seriously designing automated systems that can tell which online videos are fabricated footage of real people, or “deepfakes.” Researchers say they are "vastly overwhelmed" by the computer-generated fake video technology, which has the potential to mislead voters in the 2020 elections. - WAPO

Articles: FORTUNECNNFINANCIAL TIMESWSJ

More: The concerns surrounding deepfakes led Facebook to launch its recent deepfake detection challenge, which asks researchers to build open-source detection tools that can spot deepfake videos. The challenge includes a data set, which contains 100,000-plus videos that were specially created from scratch to help with deepfake research.

15. The AI robot Ai-Da sold $1.26 million worth of art at her first solo exhibition in June. Ai-Da, who uses AI processes and algorithms to create her original works, was on display earlier this year at the University of Oxford's St John's College. - INTERESTING ENGINEERING

16. Drive.ai shut down after Apple purchased the autonomous vehicle startup and hired many of its engineers. The startup initially pursued self-driving software systems and intelligent communications systems before turning its focus to deep learning software combined with hardware to make self-driving retrofitted kits for businesses. - TECHCRUNCH

17. The controversial DeepNude app that used GANs to "undress" women in photos was discontinued. Over the summer, the app created a backlash that forced its creators to stop selling it, saying that “the probability that people will misuse it is too high.” - MIT TECHNOLOGY REVIEW

18. SoftBank announced a second megafund of roughly $108 billion that's geared toward the acceleration of AI. The Vision Fund 2, which secured pledges from Apple, Microsoft, and others, is investing in "market-leading, tech-enabled growth companies." - CNBC

19. The Chinese AI startup Megvii, which owns Face++, filed paperwork for its initial public offering on the Hong Kong Stock Exchange. The planned IPO, which recently was dealt a setback, would make Megvii the first on a lengthy list of Chinese AI companies to go public. - REUTERS

20. In a joint debate about AI back in August, Elon Musk and Jack Ma had strikingly different views about how the technology will benefit or hurt humanity. The Tesla CEO and the former Alibaba Chairman were the highlights of the World Artificial Intelligence Conference in Shanghai. - BUSINESS INSIDER

Articles: CNBCYOUTUBEFORBESINC.COM

More: Specifically, the two vehemently disagreed on whether machines will eventually become smarter than people. Musk said AI will be "much smarter than the smartest human," but Ma argued that machines will never control humans, as "computers might be more clever" but "human beings are much smarter." Similarly, Ma disagreed with Musk that AI is a threat or "something terrible," noting that "The best resource on earth is not the clothes, not the oil, not the electricity -- it’s the human brain."

21. Ten months after first announcing the transition, Google absorbed the health division of DeepMind. The London-based AI company — already a subsidiary under Google’s parent, Alphabet — has come under scrutiny in the past over its data-sharing practices, including access to patients' health data without consent. - BBC

22. The U.S. government in October placed eight Chinese AI firms on a blacklist – prohibiting them from doing business with U.S. companies without a license – over alleged human rights violations against Muslim minorities in China’s Xinjiang region. The list includes SenseTime Group Ltd., the world’s largest AI startup, and Megvii Technology Ltd. - BLOOMBERG

Articles: BUSINESS INSIDERSCMPCNBCCAPITAL WATCH

More: Critics have argued that the U.S. blacklist could actually stimulate China's domestic AI chip manufacturing industry. China purchases about 90 percent of its AI chips from foreign companies and the blacklist may force them to find their supplies elsewhere, including at home.

23. Canadian authorities denied visas to 24 prominent AI experts from Africa and South America, preventing them from attending an industry conference in Vancouver. The situation triggered an outcry from industry experts, who argued that such denials are becoming a systemic problem and prevent diversity in the field. - CNN

24. Following criticism from groups like the ACLU, the Department of Homeland Security dropped plans to expand facial recognition checks to U.S. citizens at airports. The department currently only scans non-U.S. citizens traveling into the country, but recently said it could expand that to include all travelers. - TECHCRUNCH

25. A group of Democratic lawmakers recently sent a letter to the U.S. secretary of Housing and Urban Development (HUD), voicing concerns about the use of facial recognition technology in federally assisted housing. New York Times article reported on the rising use of face recognition to scan people in public and federal housing buildings. The lawmakers argue that the technology can infringe on basic privacy and protections and has been shown to be inherently biased against people of color, women, and non-cisgender people. - THE HILL

Written and curated by Beth Duckett, a former reporter for The Arizona Republic who wrote a book about the solar industry and frequently covers hobby and commercial drones. You can follow her tweets about the latest news in artificial intelligence here.

Subscribe to Inside AI