Sign up to Madano’s weekly summary of key developments in Artificial Intelligence here.
The week’s biggest story was Elon Musk’s OpenAI team of bots defeating human players at the video game Dota 2. It was widely hailed as a major breakthrough because of the nature of collaboration between different neural networks to defeat human players, with Bill Gates adding particular weight to the supporting chorus. The Verge’s James Vincent, however, highlighted that the bots had yet to take on the professionals.
The breakthrough brings more prominence to OpenAI, a non-profit founded on the principles of building Artificial General Intelligence (kind of AI for the purist) for public good, not harm. The name of the endeavour itself, so closely echoing open-source, has raised debate about the dangers of open approaches to AI, just as Musk and his team have highlighted the dangers of closed AI development. If much AI development and research becomes freely available, are their dangers it could be built upon by those with malevolent intent? So goes the argument, but the team at OpenAI won’t be worrying this week, instead basking in the glory of their win, even if it was against the amateurs.
News in Brief:
Matt Hancock at the Opening of Cyber Security Innovation Centre
On opening the new Cyber Security Innovation Centre, Culture and Digital secretary Matt Hancock, reemphasised the importance of AI in driving the UK economy post Brexit. As reported here by Computer Weekly’s security editor Warwick Ashford, the Government minister proclaimed that “cutting-edge technology, coupled with creative and artistic genius, is the fulcrum on which our country will be built.” Hancock also reiterated the Government’s belief that ethical AI is a potential differentiator against competitive markets and highlighted the forthcoming role of the Centre for Data Ethics and Innovation.
Hancock touched on the importance of data security given the reliance of AI upon that specific commodity. However, based on Ashford’s report, the Minister might have said more about cyber than the role AI might play in enhancing cyber defences and protecting systems and data.
Government Announces Key AI Appointments – Including DeepMind cofounder
The Government announced the appointment of Dr Demis Hassabis as an adviser to the new Office for Artificial Intelligence. In addition, CognitionX co-founder Tabitha Goldstaub was named chair of the new AI Council and the UK’s AI Business Champion, while Professor Dame Wendy Hall was confirmed as first Skills Champion for AI in the UK.
More detail here: https://www.gov.uk/government/news/world-leading-expert-demis-hassabis-to-advise-new-government-office-for-artificial-intelligence
Upcoming AI Events
Written by Darran Messem, Director, Transport & Sustainable Development
New vehicle registration data published by the UK Government this month shows a substantial year-on-year increase in hybrid-electric vehicle registrations in Quarter 1 2018, in contrast to a substantial decline in the overall market, and a decrease in pure electric vehicle registrations. The new data confirms the view that whilst there is a significant increase in the number of people considering buying an electric car, the large gap between consideration and purchase suggests that substantial barriers to adoption remain.
Hybrids up; fully-electric down
Nearly 33,000 hybrid-electric vehicles were registered in Q1 2018, according to Department for Transport data. This as an increase of nearly 15% on Q1 2017 hybrid-electric registrations of just over 28,000. In contrast, total new light-duty vehicle registrations fell by 12%, and petrol-only and diesel-only registrations fell by over 13%. Plug-in hybrid-electric registrations increased by 33% from 8,481 vehicles to 11,301, but fully-electric and range-extended electric registrations fell to 4,519 vehicles from 5,502, a decline of 18%, continuing the divergent trend between plug-in hybrid and fully-electric that’s been apparent since quarterly data was first published in 2014.
The data suggests that new vehicle buyers (and leasers) are not sufficiently persuaded by the benefits of electrification to opt for large numbers of fully-electric vehicles. This is despite the growing intention to buy an electric vehicle in the future, highlighted previously. Electric (pure and extended range) vehicles accounted for just 0.62% of total registrations in Q1 2018, compared with 0.67% in Q1 2017, despite the proportion of over-16s intending their next vehicle to be electric rising from 8% to 12% over the same period.
Rich person’s toy?
Regional uptake of Ultra Low Emission Vehicles (ULEVs) also released by the Department for Transport in June shows that the greatest increase in ULEV uptake between Q1 2018 and Q1 2017 was in relatively affluent commuter areas. The three regions showing the greatest percentage increase were Warwickshire (188%), Gloucestershire (173%) and Oxfordshire (158%). There is anecdotal evidence many EVs and ULEVs are the second cars in a household. It’s also no coincidence that two of these three counties have relatively high exposure to automotive research and manufacturing employment. Unsurprisingly the high-uptake regions have a relatively dense high-speed charging network, but this isn’t necessarily the key driver of increased uptake. The lowest rates of increase were in the North East, Greater Manchester and Lancashire; the latter two also having relatively dense high-speed charging networks.
2040: An achievable target?
The UK Government has set a 2040 target for the replacement of new conventional petrol and diesel engine vehicles with zero-emission vehicles. For this target to be met with plug-in hybrid and electric vehicles (a tougher scenario than including hybrids without plug-in capability), a substantial acceleration in uptake is required. The target is not achievable at the current rate of growth – on a ‘straight-line’ basis – of an additional 0.5 percentage-points market share per annum. A compound annual growth rate of at least 19% will be required. Application of a technology-adoption S-curve model (Rogers, E M., 1962) shows how a typical technology-adoption profile that fits with the trend to date may deliver the target, with the peak rate of growth in the curve of 8.9 percentage points (absolute change) in 2029.
Achieving this rate of increase requires current barriers to adoption to be addressed quite urgently. Given the gap between intent and actual purchase, and the gap between hybrid and pure-electric uptake, the data suggests the appeal of electrification is rising but confidence in electric vehicles has a long way to go to match confidence in internal combustion engines, particularly petrol. In the next article in this series we’ll be looking at the barriers to EV uptake, and particularly how communication can help address them.
On July 18 Madano’s Transport Practice will be hosting a breakfast meeting to discuss our analysis of UK electric vehicle uptake with key industry stakeholders. If you’d like to attend, please contact Darran Messem.
Written by Dominic Weeks, Head of Technology
There were some brilliant events hosted across the capital this week, stimulating more debate and enthusiasm for developing AI as a major part of our economy and society in the UK.
At CogX, CognitionX’s flagship event under the glorious sunshine at Tobacco Dock, there were some fascinating discussions and exciting companies showcasing their innovations. From psychographic recruitment tools like Arctic Shores that can reduce implicit bias, to SeeFashion, an analytics platform that helps retailers better understand customer intent, the show floor was full of technologies that offered substantial improvements to how we do business. Bridging the Uncanny Valley of companionship, Tony Prescott and his team from Consequential Robotics demonstrated MiRo – a companion robot development kit in canine form. Positive applications of artificial intelligence were everywhere to be seen.
The next day, the AI Summit kicked off at the ExCel with an array of big name speakers. Uber’s Chief Scientist Zoubin Ghahramani suggested a mind shift was needed to stop thinking of AI as always inspired by or replicating the human brain just because that’s where neural networks have their inspiration. It is perhaps a limiting assumption given that the human brain cannot actually store very much information at all, and it perhaps exacerbates fears of replacement. Instead, he highlighted exciting areas where AI can augment our own capabilities and experiences.
Norm Judah from Microsoft highlighted the differences between open and closed loop AI (such as autonomous vehicles), pinpointing the former as a great stream for creating augmented experiences for people, conquering “analogue hot spots” and powering digital transformation.
The Government voice was there to be heard too. Matt Hancock, Secretary of State for DCMS, used his appearance at the AI Summit to announce a new startup visa with a Dragons Den approach. He also pronounced that the current pace of change we are witnessing is the slowest we will see in our lifetime, precisely because AI is compound, not simply additive. Business secretary Greg Clark celebrated the density and close connections of our universities and tech businesses, and called for further collaboration to power AI innovation. He also asked the audience to take another look at history and reconsider the Luddites as “skilled artisans, ordinary people frightened for their future place in society” before underlining the importance of winning, maintaining and deserving public trust in AI.
The suits were not about to have the final word, however. Nesta hosted a fascinating FutureFest Forward event on Thursday evening which examined the impact of AI on art and creativity. Alphr features editor Thomas McMullen highlighted some huge leaps made with AI that are already enhancing the creativity of gaming. Yuli Levrov, Director at Reactify, questioned the concerns about human authorship, introducing a small slice of Barthesian theory into the AI discussion. UCLA professor and media artist Victoria Vesna shared a great machine learning research/installation example where humans mimicked bird song and both the system recording and the human participants learned and changed behaviour over time.
All panellists agreed on the almost limitless potential for human creativity in new and exciting areas thanks to automation, and the evolving definition of what creativity is. Vesna underlined the importance of maintaining our off grid skills and building that resilience into the education system.
It is refreshing to take a step beyond the news pages which, needs must, often play to our fears about the loss of our place in the world caused by AI development, from the erosion of our faculties as sentient beings, to the destruction of our earnings potential as workers.
From the discussions to the demonstrations, this week’s events painted a bright picture of our future living side by side with artificial intelligence. The positivity on display perhaps explains why public sentiment is shifting. An ARM survey found that 60 percent of people (globally) felt that AI will make society better, with just 22 percent feeling it would be made worse.
It would be a mistake for industry and government to rest on their laurels though. Huge barriers still exist to the implementation of certain technologies and ethical concerns were certainly a top area of debate this week. Furthermore, disruption will cause uncertainty and pain for people as the business secretary underlined, and it is critical that everything is done to reduce that friction and also balance the debate with the positives on display.
In addition, as can be seen from our weekly news in brief section, the headline news often focuses on the more terrifying aspects. The fourth estate should absolutely be keeping these questions alive and investigating troubling ethical quandaries associated with technology development. It is down to the sector to also promote the positive vision.
News in Brief
Written by Dan Townshend, Research Manager, Insights practice
Pharmaceutical companies are investing heavily in medical technology and it is easy to see why. Drug development has been declining for decades and there is a real need to find new ways of diagnosing, tracking and ultimately treating some of the more challenging diseases.
Using technology to make treatment decisions will mean that new scales for measuring indications or symptoms will need to be developed. In the past we have been limited to largely episodic, in-clinic measurement but now we can look at the kind of metrics that will help with increasingly complex neurological conditions, such as sleep patterns, number of words spoken in a day, and what proportion of the day we spend sedentary. And we can do so in a real world environment protected from the biases of the ‘lab’.
This really matters when you consider resources required to run effective clinical trials based on traditional scales collected episodically. At phase 2, a drug can appear to be acting effectively but by phase 3, it can show no improvements at all. Not only does this come at great cost to the pharmaceutical company (which ultimately affects the cost of all new drugs) but also to the thousands of patients that enter into a trial who may feel that they have wasted months or even years of their lives.
There are of course a number of challenges medical technology would have to overcome to be widespread, such as how to interpret data, what role the various actors play in this interpretation (bearing in mind we are bringing together individuals from numerous fields, including computer science, bioinformatics and health), and accessibility of the required technology. However, even if all these issues (and various others) were overcome, there would still remain one major barrier: public acceptability.
The success of existing lifestyle technologies and apps, such as FitBit, may suggest that people are increasingly accepting of the reliance on technology for making decisions about their health. However, having a vibrating watch encourage you to go for a walk is one thing, having an app tell you what treatment you should be on for your bowel disease is quite another.
How prepared would the public be to having their treatment decided by technology?
One way to approach this question would be to look at the rather timely issue of data privacy and security. According to Yuval Noah Harrari, eventually computers will “know me better than I know myself”. On the one hand, this promises to lead us to more effective, personalised medical care. On the other, there is something a bit “creepy” and unsettling about this.
If you were horrified by the recent collusion between Facebook and Cambridge Analytica and your social media data, you would be forgiven for having similar concerns about your health data. After all, it is far from clear who would have access, legitimately or illegitimately, and who would ultimately own it. In a recent PwC survey, 80% of respondents felt that healthcare apps should not have access to patient records. Clearly, there is some way to go in establishing trust.
That said, the emergence of GDPR does point the way to how we can embrace medical technology, if we can apply the same principles to the protection of personal health data. If governments take a lead in articulating the benefits of such technology powerfully (by pointing to the potential impacts to national healthcare systems, industry and, most importantly, to the patient) and couple this with robust safeguards people may be more accepting of it. After all, there are already organisations that have access to our personal medical records when they need them, such as insurance companies, and researchers also have access to our anonymised health data, undoubtedly to our collective benefit.
So long as data is secure, transparent and is ultimately controllable for the user, there is every reason to believe the public will come to accept medical technology and benefit from its many promises. All we need to do is to get the appropriate frameworks in place to allow it to truly flourish.
Written by Evan Byrne, Senior Account Executive, Energy Practice
Green financing is a hot topic right now. Last month the House of Commons Environmental Audit Committee (EAC) published its report Green finance: mobilising investment in clean energy and sustainable development, which argues that green investment has plummeted, driven largely by poor policy decisions and a lack of consistent direction from Government.
Figures from Bloomberg New Energy Finance show that in cash terms new investments into low carbon energy fell by 56 per cent in 2017, compared with 2016, which itself saw a drop of ten per cent from the year before.
This has led many commentators to sound the klaxon. Some claim that the fall in green investment will in turn prevent us from successfully tackling climate change down the road. It has led to calls for a complete overhaul of how we manage green financing with suggestions ranging from calls for the UK to introduce a Sovereign Green Bond, to the Bank of England building climate-related risks into its macroeconomic models. The Bank of England also recently told investors to back green companies or face the risk of huge losses from climate change.
These measures may be useful. There is a strong case to be made for reviewing how low carbon technologies are funded going forward, as policies and mechanisms introduced over the past decade may now no longer be fit for purpose. Several of the recommendations from the EAC report do in fact provide innovative solutions that could lead to more efficient and effective green financing.
Are green bonds actually ‘green’?
There are indeed some issues around ensuring that ‘green bonds’ are in fact green (rather than potentially the result of greenwash strategies), and a UK sovereign bond could help set a standard by which private bonds could be benchmarked.
Another recommendation from the EAC report is for the introduction of a carbon tax that increases over the long term, as generators and suppliers would have to adopt low carbon technologies in order to avoid future costs.
Renewables becoming ever more affordable
However, one might question whether a fall in green investment in cash terms is a regressive step. There is ample evidence showing that the capital costs associated with renewables in particular have fallen to record low levels. Of particular note is the fact that renewables and other low carbon generation’s share of the energy mix continues to rise.
All of this in fact suggests in terms of outputs – generation, capacity, total number of projects under development – that renewable and low carbon technologies continue to perform well and that the “fall in investment” is reflective of the vastly reduced costs associated with these technologies. There is consistent global data which shows that generation costs continue to fall, driven by global competition. In the UK, renewable projects are now being developed which are completely subsidy free, which shows how the scale of change these technologies have undergone financially.
The EAC Report proposes several valuable solutions, which the Government should look at implementing to improve the green financing space, and to continue the trend of enabling cheaper and cheaper renewable and low carbon energy.
However, the Government must be careful and not misunderstand the data it has been presented with: the “fall in investment” has been presented in cash terms. It is easy to argue that any fall in cash influx is a backwards step, but this argument fails to account for new efficiencies creating savings or fall in initial capital costs.
The Government must ensure that any future policies around green finance tackle the problems we will face in the future, and not revert to old policy positions which are not fit for purpose in today’s energy landscape.