Latest Artificial Intelligence News
Hailed as the most significant test of AI since machine triumphed over man in chess nearly 20 years ago, Google’s AlphaGo has beaten Lee Sedol in the ancient Chinese board game Go. Described as the ‘Holy Grail’ of AI challenges, the 2,500-year-old game is exponentially more complex than chess and requires, at least among humans, an added degree of intuition, making last week’s results all the more intriguing. AlphaGo’s victory marks a major development in reinforcement learning, the ability for a machine to play against itself and adjust its own neural networks based on trial and error.
This milestone in AI development, though notable in its own right, holds even greater significance when considering its real-world applications. Google DeepMind, AlphaGo’s creator, is currently developing a software partnership with NHS hospitals to alert staff to patients at risk of kidney failure. Though the app at present does not utilize machine intelligence, working as a resource allocator to collate patient information and provide record of what actions have and need to be taken, the NHS’s partnership with one of the world’s leading AI companies has far reaching implications for the future development of smart diagnostic agents.
Other emerging real-world applications of AI include virtual assistants to develop a deeper understanding of what you are trying to accomplish in business, or in your social life. Though there is still far to go in terms of developing intuitive and flexible response – or “making smart assistants actually smart”, according to DeepMind founder Demis Hassabis – current virtual assistants such as x.ai uses AI to schedule meetings “with intelligence and human empathy”. However, a study published this week looking at virtual assistants’ ability to respond in crisis, found that while Google and Apple’s assistants may be great for finding the nearest gas station, they fall short in times of distress, unable to respond sensitively or make referrals to helplines.
Important AI news also came last week in the realm of driverless cars. The Department of Transport announced its plans to trial driverless lorries in the UK and notably, after several years and more than 1.4 million miles of test driving, one of Google’s self-driving Lexus SUVs was involved in an accident for the first time. Though no one was injured, the event is raising important questions over how we will weigh and judge 100% autonomous cars against a track record that cannot 100% guarantee our safety.
This week has also seen the playgrounds for AI experimentation widen – infinitely – with Microsoft announcing that Minecraft will allow computer scientists to evaluate and develop AI software using its virtual landscapes. In addition to cutting costs involved in AI research simulation, Microsoft sees the game’s openness and creativity as presenting a unique opportunity for AI agents to learn from the environment and eventually collaborate with humans within the virtual space.
Though still speculative, experts believe that developments of tech and AI agents within Minecraft’s virtual reality will be able to inform robotics in the real word.
Artificial Intelligence at home
Companies developing increasingly connected, autonomous and intelligent machines are making bold promises about future products and services, but are we, to use the Gartner Hype Cycle, climbing the “slope of enlightenment” for these new technologies, or are we sat precariously on the “peak of inflated expectation”?
We explore important developments surrounding smart homes, virtual assistants and artificial intelligence, and what they mean for industry players and consumers.
Smart homes are still a little bit dumb
So what is going wrong with smart home products? Most of us are still far away from a Jetsons-like future where our central heating, lighting and music systems learn exactly how warm, awake and happy/sad we want to be at any time of day.
Nest was meant to be the “trailblazer that led us to the smart home revolution”, Washington Post proclaims, but uptake of products has been slow to the point of failure. Last month, for example, much to the chagrin of its customers, Nest closed the cloud platform powering its Revolv smart home business, turning each of the Revolv hubs into £210 lumps of plastic.
Growth in the connected homes market is being limited by various issues affecting the consumer experience, noted in a recent survey by Accenture, such as high prices, confusing value propositions, concerns over privacy and security (including security locking out actual users), buggy software, arduous set-up and login processes, and a general lack of operability between other systems. When one combines these issues with the possibility that any device you buy could be made redundant shortly afterwards… the future for standalone smart home products does not look very bright. Thus, despite the high levels of hype surrounding the Internet of Things throughout 2015, consumer interest in the connected homes market is waning (Argus consumer research).
Good morning Sir. May I offer you some coffee?
However, with the large tech players – Amazon, Apple and Google – all working on intelligent voice assistants which could sit more comfortably on our phones and follow us everywhere, WashPo notes, there may be little use for separate hubs in homes of the future. Onstage at the Code Conference last week, Recode reports, Amazon CEO Jeff Bezos affirmed it had over a thousand people working on its voice assistant offering, Alexa and Echo. The statement of intent comes shortly after Google announced their own ecosystem, Google Home, at their I/O last month and rumours spread of Apple working on its own hub for Siri. Although the Amazon Echo is another standalone device, it is important to view it as part of an ecosystem; Bezos argued that the race for this market lies in the ability of its players to offer cross-device apps and services, mirrored by Apple’s decision to open up Siri to external app developers.
The recent Mary Meeker Report predicts that by 2020 at least 50 percent of all searches will be through images or speech, so it is clear why Google parent Alphabet is interested in the sector. Google is certainly the head of the pack with artificial intelligence, as shown in its recent successes with AlphaGo, driven in part by the enormous amounts of data created from its core services. These developments around machine learning, in theory, may lead to more compelling assistant products.
However, Stratechery asks, will being the best technologically be enough for voice assistants? Google’s initial success, the blog post states, was largely because their search product was so superior to alternatives, and the cost of usage so low (as in typing in an URL). The cost of usage with voice assistants, on the other hand, is far greater, particularly when they require a standalone device. Google’s recurrent challenge, the article concludes, is overcoming the friction involved with distributing the service on other devices, namely the iPhone. Google Maps has been widely regarded as better than Apple Maps, for example, but is used three times less than Apple Maps on iPhones.
Before Google and Apple launch their rival products to Amazon’s Echo, it remains unclear as to whether one of the large players will be able to offer a truly differentiated ecosystem. It is clear, as concluded by recent consumer research by Creative Strategies, that voice-user interfacing is moving towards the mainstream, with consumers increasingly looking to these services for value and convenience. It may be some time, however, before we’re all using them in public.