Inside | Real news, curated by real humans
Inside AI

Inside AI (Aug 14th, 2019)

1. Several current and former employees of SoftBank-backed Engineer.ai claim that the startup exaggerates its tech expertise and AI capabilities. The news was first reported by The Wall Street Journal, which notes that the accusations "reflect a growing challenge in the tech world of assessing a company’s proficiency in artificial intelligence." Last year, Engineer.ai raised Series A seed funding of $29.5 million, led by Lakestar and Jungle Ventures. The startup uses a human-assisted AI to help companies build custom-made software. - WSJ

  • Email gray
  • Permalink gray

2. AI researchers recently launched SuperGLUE, a series of benchmark tasks that measure the performance of language-understanding AI. SuperGLUE, which uses BERT as a model performance baseline, comes from Facebook AI Research, Google’s DeepMind, University of Washington, and New York University. It's meant to be more complicated than its predecessor GLUE (General Language Understanding Evaluation), which debuted last April via a partnership between NYU, the University of Washington, and DeepMind. In a blog post, Facebook AI researchers said SuperGLUE has "new ways to test creative approaches on a range of difficult NLP tasks" that are focused on core areas like "sample-efficient, transfer, multitask, and self-supervised learning." - VENTURE BEAT

  • Email gray
  • Permalink gray

3. Amazon's facial recognition technology Rekognition can now detect “fear” from reading a person’s face. The Rekognition software offers a comprehensive range of tools for face detection, analysis and recognition in images and videos. Engineers have improved its accuracy for detecting the seven main emotions: happy, sad, angry, surprised, disgusted, calm and confused. Now, the company is adding fear to that list. It has also made updates to the facial analysis tool and improved its accuracy in identifying genders. What's more, Amazon updated the software last week to be able to detect violent content like blood, wounds, weapons, self-injury and corpses. - GEEKWIRE

This story first appeared in today's edition of Inside Amazon.

  • Email gray
  • Permalink gray

4. In an opinion piece for Wired, Robust.AI's Gary Marcus looks at the deeper meaning behind DeepMind's losses of more than $1 billion during the past three years. The Alphabet-owned company lost $154 million in 2016, $341 million in 2017, $572 million in 2018, raising several questions: "Is DeepMind on the right track scientifically? Are investments of this magnitude sound from Alphabet’s perspective? And how will the losses affect AI in general?" DeepMind is also more than $1 billion in debt, which is due over the next year. Despite the financial challenges, Marcus notes that losses are expected, given the large amount of research that DeepMind is doing every year, "perhaps more than in any previous AI research operation, but far from unprecedented when compared with the sums spent in some of science’s largest projects," he adds. - WIRED

  • Email gray
  • Permalink gray

5. Researchers used AI techqniues to determine whether butterfly species can co-evolve similar patterns on their wings, a phenomenon known as Müllerian mimicry. The theory, from naturalist Fritz Müller, proposed that species can mimic each other for mutual benefit. The researchers, from the University of Cambridge, the Tokyo Institute of Technology, the University of Essex, and the Natural History Museum London, trained a machine-learning algorithm to measure the visual variations between different subspecies of Heliconius butterflies. According to Phys.org, scientists found that different species "act both as model and as mimic, 'borrowing' features from each other and even generating new patterns." - PHYS.ORG

  • Email gray
  • Permalink gray

6. The AI field continues to have a dearth of women - which has the potential to skew data further. Fewer than 20 percent of the researchers applying to AI conferences are women, and only 25 percent of undergrads studying AI at Berkeley and Stanford are female, according to the AI Now Institute's report "Discriminating Systems: Gender, Race and Power in AI," which you can find here. In addition, Nesta's report on "Gender Diversity in AI Research" found that only 13.8 percent of AI authors in the pre-prints website arXiv are women. - AI NOW INSTITUTE

  • Email gray
  • Permalink gray

7. The Atlantic League of Professional Baseball will begin testing an AI-based ball-tracking system, which will be observed by Major League Baseball for potential use as a robot umpire. Radar-based Trackman can detect and analyze velocity, movement, pitch type, and other details about a ball as it enters the strike zone. - CNET

  • Email gray
  • Permalink gray

8. More companies, cities, and organizations are using AI to fight ransomware attacks, according to The Wall Street Journal. A Capgemini study found that about 75 percent of the organizations polled are testing AI technology for cybersecurity purposes. One such example is the city of Las Vegas, which has been using AI augmentations over the last three years to detect and respond to malware threats 24/7. - WSJ

  • Email gray
  • Permalink gray

9. Forbes interviewed its Coaches Council about the downsides of using AI to hire new employees. While more companies are using AI systems in the hiring process, it can be off-putting for job candidates and lacks a "gut instinct." AI also has the potential to ignore resumes that aren't SEO-optimized, and fail to predict soft skills, tone, and personality, according to members of the council. - FORBES

  • Email gray
  • Permalink gray

10. Musician and actor Ice-T voices an AI in the upcoming Borderlands 3 video game. The legendary emcee, who is said to be a big fan of the Borderlands series, plays the character Balex, an AI that's trapped inside the body of a teddy bear. The game is due out on September 13 for PS4, Xbox One, and PC. - GAMESPOT

  • Email gray
  • Permalink gray

Written and curated by Beth Duckett in Orange County. Beth is a former reporter for The Arizona Republic who has written for USA Today, Get Out magazine and other publications. Follow her tweets about breaking news and other topics in southern California here.

Editor: Kim Lyons (Pittsburgh-based journalist and managing editor at Inside).

  • Email gray
  • Permalink gray

Subscribe to Inside AI