Research from the University of Plymouth found that, while adults are not susceptible to bad advice from robots, children are easily led astray by machines. The study aped the famous Asch paradigm, which has long been used to examine how susceptible people are to peer pressure. In this case, however, the human peer was replaced by a robot.
The findings post yet another dystopian image of the AI future and freshly surface the importance of establishing clear ethical frameworks in AI R&D. This is an area that the UK Government is dialled in on, in terms of both protecting the public at large but also in forging a commercial niche for the UK as a centre for ethical innovation in AI. Technology innovators and influencers need to come together with government in a bid to have early discussions on how abuses and side-effects of AI innovation can be prevented.
Machine-learning system can identify 50 types of disease better than leading experts
A new machine-learning system developed by DeepMind and Moorfields eye hospital is as good as the world’s best experts at detecting eye problems, reports Samuel Gibbs in The Guardian. The development is a particularly exciting one due to the transparency it provides for checking its assessment. “The maps and any differing or ambiguous results can be shown visually to a clinician for their own interpretation and explanation of the referral result,” writes Gibbs. This is a refreshing change from many of the “black box” machine learning algorithms that are bemoaned for not allowing humans to check and understand how the systems arrived at a particular conclusion. Clinical trials and regulatory approvals await, but this breakthrough is seen as an answer to the increasing lack of available experts.
Divvin’ be nebby Alexa
One major fear relating to the explosion in virtual home assistants is that they are snooping on our conversations. Less thought has been given to how they subtly might mould our own communications choices. According to research by Newcastle’s Life Science Centre, home assistants threaten regional accents as locals shift their speech closer to received pronunciation to appease their smart assistants. After all, nobody wants to get caught in a voice-recognition powered lift in Scotland.
News in Brief:
- Facebook testing Artificial Intelligence system for flagging hateful content in Myanmar: https://www.washingtonpost.com/technology/2018/08/16/facebook-turns-artificial-intelligence-fight-hate-misinformation-myanmar/?noredirect=on&utm_term=.54e7017d611e
- U.S. spending cements AI in America’s approach to defence: https://www.fastcompany.com/90219751/pentagons-artificial-intelligence-programs-get-huge-boost-in-defense-budget
- One third of businesses believe that artificial intelligence will be widespread in the enterprise by 2028: https://www.v3.co.uk/v3-uk/news/3061237/most-businesses-will-use-artificial-intelligence-by-2028
- Human intervention is necessary to adjudicate the bias in the programmer, the context and the language itself, according to a post published by Chatham House: https://www.chathamhouse.org/expert/comment/human-factor-essential-eliminating-bias-artificial-intelligence
- Berlin’s Taxfix closed a $13 million Series A funding round, plans to invest heavily in machine learning to improve its mobile tax filing app: https://techcrunch.com/2018/08/09/taxfix/
- Thames Valley is a hotspot for potential AI innovation according to the Times: https://www.thetimes.co.uk/article/heathrows-backyard-has-all-the-right-connections-r5c2whhb5
- FiveAI gets permission to conduct large scale study of road behaviour in order to build safer driverless cars: https://www.telegraph.co.uk/technology/2018/08/14/british-start-up-take-road-data-gathering-cars-streets-london/
Food, glorious food
Innovate UK invited entries for a £20 million competition for research projects that represent a crossover between the agri-food sector and real-time robotics sensing, data, AI and earth observation. The initiative is part of the Government’s broader £90 million initiative to grow more food to support a swelling population. More information: https://www.gov.uk/government/news/efficient-and-sustainable-agriculture-apply-for-funding
Contrasting assessments of DCMS ministers’ savvy
In other news, ad trade mag Campaign issues a withering appraisal of new DCMS Secretary Jeremy Wright, referencing that he has but twice uttered the words digital in the Commons – https://www.campaignlive.co.uk/article/lawman-cometh-will-jeremy-wright-good-advertising-industry/1490004
However, MP Margot James, Minister for Digital and the Creative Industries, was shortlisted by Computer Weekly as one of the most influential women in IT https://www.computerweekly.com/news/252444988/Voting-open-vote-now-for-the-most-influential-woman-in-UK-technology-2018?Offer=maxitwittervoting
Upcoming AI Events
- Introduction to Machine Learning – Cambridge – Friday, August 3, 2018 7:00 PM https://www.meetup.com/Cambridge-Machine-Learning-Meetup/
- Chat & Voice Assistant Meetup – London – Tuesday, August 28, 2018 6:30 PM to 9:00 PM https://www.meetup.com/Messaging-Bots-London/events/253369172/