1 – Who’s Better At Phishing Twitter, Me Or Artificial Intelligence?

Forbes Reporter Thomas Fox-Brewster went head-to-head with a double-teamed AI to see who was the better phisher on Twitter – a human or an AI. The AI software, called SNAP-R (Social Media Automated Phishing and Reconnaissance), was created by data scientists John Seymour and Phillip Tully, both with security company ZeroFOX. Though Fox-Brewster ended up with more followers for his Twitter characters, SNAP-R outperformed his phishing attempts by 226 victims, sending out far more tweets in the two-hour match and scoring a higher conversion rate (275 click-throughs for machine versus 49 click-throughs for humans). The results point to the challenge by security staff in identifying machines versus humans, and this article comes as DARPA’s Cyber Grand Challenge approaches in the beginning of August, when two AI machines will try and out-hack each other for the first time. Seymour and Tully believe this type of AI will be freely avilable within five years, indicating a strong need to develop systems that protect against ‘malicious’, smarter AI.

(Read the full article on Forbes)

2 – Researcher Proposes Social Emotions Test for Artificial Intelligence

At the 2016 Annual International Conference on Biologically Inspired Cognitive Architectures (BICA) in New York, Researcher Alexei Samsonovich of the Moscow Engineering Physics Institute presented his ideas on developing and testing an artificial intelligence that experiences human-level emotions, an important component in conscious awareness. In Samsonovich’s vision, the AI would be tested against a human in controlling virtual players in a computer program that involves emotionally-laden social interactions (such as subordination or mutual trust). A Turing test would be used to assess whether a player is human or machine, with the determining factor being ‘machine preservation’ (i.e. the machine manifests behaviors that show it trying to rescue the machine over the human in a given situation). Samsonovich announced that he plans to begin working over the next 18 months on a virtual being that has goal-setting capabilities, can establish social relationships with humans, among other qualities of emotional and narrative intelligence.

(Read the full article on Phys.org)

3 – Kickstarter Campaign Launched for World’s First AI-written Feature Film

Artificial intelligence is being recruited to help write a successful feature film, with a Kickstarter campaign to do so launched this week. So far, 94 backers have brought in a total of $5,538 (out of a requested $30,000) with 57 days to go and a planned project completion date of October 2017. The film Impossible Things will be a collaboration between human and AI software that will use data collected about features of other successful horror flicks to help write engaging plot points. Jack Zhang, Founder of the data analysis company Greenlight Essentials, is spearheading the project, already in development for five years. Zhang’s target audience is women under the age of 25, a factor the the software included when cranking out features that sell well with that particular crowd. While the effort isn’t the first partly AI-engineered film (Sunspring was the first), it’s a further step in the direction of AI collaboration in art.

“Sunspring”, the first AI-created screenplay.

(Read the full article on The Guardian)

4 – Artificial intelligence Identifies Bat Species that could Harbor Ebola

Researchers studying bat species as carriers of the Ebola virus used a machine learning algorithm to help pinpoint the most likely carrier culprits. The results could help scientists better survey and prevent outbreaks of Ebola and other filoviruses in the future. Barbara A. Han, PhD and disease ecologist at the Cary Institute of Ecosystem Studies, set out with her colleagues to identify any evidence, including physiological and ecological traits, that make a particular species more likely to carry the diseases. They found that bats that covered a wider geographic range and came in contact with more mammals (along with other important variable, such as earlier maturation and higher reproduction rates) were more likely to test positive, and the algorithm used to distinguish between Ebola-free and Ebola-positive bats had an 87 percent accuracy. One of Han’s colleagues, Dr. David T.S. Hayman, stated:

“This model allows us to move beyond our own biases and find patterns in the data that only a machine can.”

(Read the full article on Healio and original research at PLoS)

5 – NYU Tandon, FF Venture Capital Announce AI NexusLab to Launch Artificial Intelligence Startups

The NYU Tandon School of Engineering is partnering with ff Venture Capital (ffVC) to launch New York City’s first collaborative academia and venture capital supported program for artificial intelligence startups. The NYU/ffVC AI NexusLab program will recruit five startups from around the world whose  missions fall into one of five industries: software, data and security, health care, finance, and media and publishing. John Frankel, ffVC founding partner, said:

“We’re excited to be partnering with an institution as esteemed as NYU on a program that we believe can help guide New York City to become the artificial intelligence center of excellence…We’re hopeful that the companies entering the program will harness the combined resources of NYU and ffVC to amplify their growth and help shape the future of technology.”

Different from an accelerator, the AI NexusLab is designed for integration into and support of the NYC community, with one measure being a public technology conference planned at the conclusion of the program in April, providing networking opportunities for entrepreneurs, academics, technologists, and others.

(Read the full article on NYU Tandon School of Engineering News)

Image credit: allevents.in