I will be chairing the Keynote Theatre on Day 1 of Big Data & AI World in March at the ExCeL Exhibition Centre in London. on Day 2 he will be part of a panel on Putting AI into Practice and also holding a keynote on responsible systems design.
bertie
Mitsuku wins 2019 Loebner Prize and Best Overall Chatbot at AISB X
For the fourth consecutive year, Steve Worswick’s Mitsuku has won the Loebner Prize for the most humanlike chatbot entry to the contest. This is the fifth time that Steve has won the Loebner Prize. The Loebner Prize is the world’s longest running Turing-Test competition and has been organised by AISB, the world’s oldest AI society, since 2014. For the first time this year, the chatbot contest was embedded in a public-outreach event AISBX: Creativity Meets Economy, that was held at the Computational Foundry on Swansea University’s Bay Campus from 12-15 September and attracted over 300 visitors.
The event combined workshops on chatbots for over 200 school children from 6 schools in South Wales with a public art exhibition, a chatbot exhibition, and a work programme on conversational AI systems attended by an international audience from the USA, Jersey, and the UK. The chatbot exhibition showed 17 conversational AIs by developers from countries such as Switzerland, Vietnam, USA, The Netherlands, Poland, UK, Jersey, Italy, and Spain. The art exhibition showed fascinating pieces and installations from international artists John Gerrard, Gene Kogan, Daniel Berio, Simon Colton, Cuan McMurrough, and Disnovation.org. From digital graffiti, synthesised news headlines, and thought-provoking works on climate and embodiment, the exhibition achieved its aim of instigating discussions amongst the audience and the organisers of the event that was co-funded by CHERISH.DE and AISB.
The longest running Turing Test competition is on at Bletchley Park.
This year the Loebner prize will take place on Saturday 8 September from 1pm until 4pm.The first 4 chatbots from the selection round will compete in the finals at Bletchley Park in Learning Rooms 3/4.
The Importance of Diversity
I will be at the Business AI and Robotics event in Helsinki in October.
Press Release on the House of Lords Select Committee Report on AI
PRESS RELEASE
The Society for the Study of Artificial Intelligence and Simulation of Behaviour (AISB) welcomes the comprehensive report by the House of Lords Select Committee on AI published on 16 April 2018 following a consultation process in the autumn of 2017 to which AISB submitted written evidence. AISB fully supports the key messages conveyed by the principles for an AI Code postulated therein and agrees that AI should be used for the common good and benefit of humanity (Principle 1). AISB would like to emphasise Principle 2 requiring explainable or intelligible AI be made compulsory for AI-based decision support in certain critical areas, as stated in #94 of the report. Principle 5, restricting AI’s autonomous power to hurt, destroy or deceive human beings also finds our full support.
However, some phrases in the principles published as #147 in the report ought to be modified to acknowledge that AI technologies are tools developed by humans. We don’t ‘work alongside AI’, we use AI to achieve certain goals or outcomes, just as with any other human-made tool. Granting AI an independent human-like existence, even through casual use of language, sets us on a dangerous course towards machines becoming moral patients; things to which we owe some moral duty. For example, Principle 2 should be amended to read ‘AI should be designed to operate on principles …’ to clarify that people are responsible for the design and operation of AI systems.
Also, Principles 3 and 4 warrant some comment:
AI cannot legally be used to diminish rights or privacy if applied according to existing or future data protection regulations and other legislation as long as there is human responsibility for the AI. We do not recommend granting any AI human rights (and responsibilities), not least since this will inevitably open up legal loopholes.
A right to education in the sense of Principle 4 needs to be aided by a counterpart stating certain restrictions to AI. A right for humans to receive the education to flourish economically without further explanation is not helpful, since often the use of AI is economically motivated. It is important that humans should use AI-based tools to tackle important problems as efficiently as possible while having been given the skills to enable them to remain economically sustainable individuals. It is important to point out that the possibilities of flourishing mentally, and emotionally refer to human attributes that should not be hindered by any AI tool, nor should education of humans be lessened by any constraints imposed by (the intent to use) artificial intelligence.
Dr Bertie Müller (AISB Chair), 16 April 2018
http://aisb.org.uk/news/174-lordsai
The Select-Committee report can be found here (HTML): Report of Session 2017-19 – AI in the UK: ready, willing and able? or as PDF: PDF version Report of Session 2017-19 – AI in the UK: ready, willing and able? ( PDF )
AI Europe 2017
Engaging conversations and good feedback on my presentation on the need for responsible and ethical AI working for humanity at AI Europe on 20/21 November 2017.
Panelist at TFM 2017
On 27 September I will be one of the panelists at the Technology for Marketing conference held at the Olympia in London. We will be discussing the role of AI in marketing. Here are the details the panel:
Panellists:
- Jeremy Waite, Evangelist, IBM Watson
- Parry Malm, CEO, Phrasee
- Tom Smith, Senior Manager, Product Marketing EMEA, Salesforce
- Berndt Müller, Chair, AISB (Society for the Study of Artificial Intelligence and Simulation of Behaviour)
Moderated by:
Roland van Breukelen, Director Customer Engagement & Commerce, SAP Hybris
Loebner Prize 2015
On 19 September the 25th annual Loebner Prize in Artificial Intelligence was held at Bletchley Park. Dr Bertie Müller, Senior Lecture in Computing at the University of South Wales and Chairman of the Society for the Study of Artificial Intelligence and Simulation of Behaviour (AISB), organised this event with the help of further AISB-Committee members for the second time running. Bertie was interviewed by BBC News (broadcast live at 10:28 on 19 September) from Bletchley Park talking about the competition and how the Turing Test was relevant to us more than 60 years after its first publication. The interview and further BBC coverage was part of the week-long Intelligent Machines season across the BBC TV and radio channels. Throughout the day further coverage of the event was provided by the BBC News channel, Sky News, and CBS.
Some of the BBC coverage of the event is archived here and in related posts.
Furthermore, Bertie is quoted in the New Scientist article entitled “Forget the Turing test – there are better ways of judging AI” suggesting alternatives to the Turing Test as a measure of (machine) intelligence.
None of the chatbots managed to fool any of the judges. The prize for the most human-like machine went to Rose (developed by Bruce Wilcox).
Finalists
Steve Worswick
Brian Rigsby
Bruce Wilcox
Ken Hurturbise
Judges
Jacob Aaron – Physical sciences reporter for New Scientist
Rory Callan – Jones
Technology correspondent for the BBC
Brett Marty
– Film Director and Photographer
Ariadne Tampion – Independent Writer and Thinker
Confederates
Paul Beesley
– Software Engineer at ARM
Emily Jones
– Admissions registrar at Moorlands School, Leeds
Paul Sobek
– Software Engineer at Imagination Technologies
Chris Wignall –
Senior Software Developer at Lotus F1 Team
Programming
What has happened to the perception of programming being a science, an engineering discipline, or even art? Today’s students seem to think programming is nothing but copying code snippets from some dodgy sources, renaming variables (if they can even be bothered with that) and hoping that by magic this conglomeration of code odds and ends will produce the desired result. This approach is certainly not science or engineering. One might see an artistic spirit in it, but to me it mostly resembles guessing, gambling, and hoping that the parser will highlight & correct some syntax errors … and when the code compiles it surely has to be correct.