artificial intelligence innovation report - deloitte
TRANSCRIPT
Artificial IntelligenceInnovation Report
Powered by2016
Artificial IntelligenceInnovation Report 2016
IntroductionOnce the preserve of science fiction, Artificial Intelligence (AI) is now firmly embedded in our day-to-day lives. Maps that compute the optimal driving route for a journey based on current traffic, smart song suggestions from Pandora, friend recommendations, job recommendations, product recommendations based on your purchase history — all are powered by AI. Even email spam filters, which learn to differentiate between email you do and don’t want to receive, rely on a form of AI.
In recent years we’ve become accustomed to seeing AI referenced in business tag lines and popular culture. In 2011 IBM Watson defeated the world’s two best players of the Jeopardy quiz show. In March 2016 Google DeepMind’s
AlphaGo eclipsed the 1997 achievement’s of IBM’s Deep Blue when it defeated Lee Sedol in a five-game match of Go. Siri and Cortana are AI assistants we can carry in our pockets. And yet these achievements only hint at the future of AI.
To understand the current state of AI and what the future holds, we must first split our understanding of AI into three definitions:
Artificial Narrow Intelligence (ANI): Sometimes referred to as Weak AI, this is the most limited form of AI and this is what is currently possible in 2016. ANI is capable of carrying out specific tasks brilliantly, using a combination of advanced algorithms, deep learning and various other techniques depending on the use. Often,
Natural Language processing is employed to facilitate human interaction.
This AI is narrow because it is incapable of carrying out any task outside of its specific purpose. Impressive as AlphaGo may be at playing Go, it would be flummoxed if you asked it for restaurant recommendations.
Artificial General Intelligence (AGI): Sometimes referred to as Strong AI or Human-Level AI. At this level, AI would be capable of performing any task to the standard of a human. Precisely when AGI will be achieved is a subject of debate, with most AI specialists predicting it will occur around 2040.
To reach AGI requires hardware comparable to the human brain, which can process
AI
Powered by
2
Artificial IntelligenceInnovation Report 2016
Introductionapproximately 10 quadrillion calculations per second on just 20 watts of power. Such a computer actually already exists in China’s Tianhe-2, which can handle 34 quadrillion calculations per second. Spread across 720 square meters, costing USD 390 million, and requiring 24 megawatts of power, Tianhe-2 is clearly inappropriate for the consumer market. However, it does demonstrate that hardware is unlikely to be the biggest block on the development of AGI.
Developing software that is as advanced as the human mind poses a greater challenge. While such software does not currently exist, the majority of computer scientists believe the most promising approach could be to build an AI that can code changes into itself — enabling it
to develop improvements on its own AI. This would enable it to become progressively smarter, and to constantly build upon its own intelligence with an exponential increase in gains.
Artificial Superintelligence (ASI): This is an AI that is more advanced than human intelligence. Due to the Law of Accelerating Returns, this state could potentially be realized in mere days of even hours after AGI is achieved.
The consequences of developing ASI are, by its very definition, unknown. This level of intelligence would soon be many billion times more intelligent than the smartest human. Because of this, it is impossible for anyone to predict with certainty what this leap in technology will bring forth. An
end to global warming, a cure to cancer, human extinction, the end of human mortality, and eternal agelessness are all being considered as very possible outcomes. Stephen Hawking describes the advent of ASI as a moment that “could spell the end of the human race”, while Ray Kurzweil believes it could introduce a human-AI synthesis or technological singularity that can best be described as a form of utopia.
This may sound like science fiction, but if we look at the ANI of 2016 we can already see the impact of these developments, most noticeably on the workforce.
It is estimated that 50% of today’s jobs across the globe will be serviced by AI in 2050. Many of the ANI examples
AI
Powered by
3
Artificial IntelligenceInnovation Report 2016
Introductionbelow already demonstrate this movement, with repetitive, data-driven, non-social occupations being the most susceptible to early replacement. Roles that require lateral, creative thinking, empathy or social sensitivity are the hardest to automate. ANI is already advanced enough to handle a wide spectrum of jobs, from creating nutritious vegan “meat” to predicting the success of a startup.
For approximately the next 25 years, ANI innovations will be restricted to carrying out specific tasks, or a defined group of specific tasks. Processing jobs considered complex to the human brain — such as advanced mathematical calculations — will be easy work for an ANI. Conversely, it’s often the most mundane tasks — for example, differentiating between
a picture of a dog and a cat— that remain incredibly complex for our current generation of ANI. As computer scientist Donald Knuth explains, “AI has by now succeeded in doing essentially everything that requires ‘thinking’ but has failed to do most of what people and animals do ‘without thinking.’”
Until AGI is achieved, ANI is reliant on machine learning algorithms which require huge data sets and an adaptation of our natural world to be capable of comprehension. In the words of John Martinis, a quantum physicist currently working at Google, “machine-learning algorithms are really kind of stupid. They need so many examples to learn.” The ANI examples below are only able to perform their specific tasks because humans have already
adapted the natural environment for them. X.ai may be able to take the pain out of scheduling meetings, but it can only do so if everyone keeps their digital calendar up to date and attaches the necessary information.
Until AGI or near-AGI is achieved this adaptation of environments will remain a necessity. But as is evident in McCann Japan’s recent appointment of an AI as Creative Director, this gap between ANI and AGI is closing fast.
CHRIS KREINCZES
EDITOR & CREATIVEDIRECTOR, SPRINGWISE
AI
Powered by
4
Powered by
We’ve seen how autonomous drones could be used to protect human populations from mosquito-borne outbreaks, and now artificial intelligence is predicting where human epidemics can occur. AIME (Artificial Intelligence in Medical Epidemiology) has developed a Dengue Outbreak Prediction platform that predicts outbreaks of the potentially fatal disease three months in advance.
The AI algorithm combines epidemiological research with other data known to impact the spread of insect-borne disease, such as population density, wind
speed and direction, rain volume, housing types, and more. It then calculates a percentage of outbreak probability in a given area. The platform determines the epicenter of a spreading disease within a 400 meter radius, and this will be the crucial area that health officials look to contain and monitor.
Having successfully trialled their platform with 87 percent accuracy in dengue fever outbreaks, the team hopes to target other highly contagious diseases, such as Ebola or the Zika virus. Could AI algorithms be used to predict other scenarios of
concern to public safety, such as riots?
AI can predict where epidemics will happenAIME use big data and AI algorithms to pinpoint the geolocation of dengue fever outbreaks, three months before they occur.
5
At a Glance
WHATAIME use big data and AI algorithms to pinpoint the geolocation of dengue fever outbreaks, three months before
WHOAIME
WHEREUnited States
Powered by
Unreel Entertainment uses AI bot Bumblebee to trawl the internet and social media for the most popular videos, making it easier for viewers to find and share new content. Now, with Unreel.me, everything a user needs to create a bespoke video streaming site and app, and build an audience, is provided in one place.
Each Unreel.me site and app uses content uploaded directly to the Unreel platform, and syncs with videos posted on social networks. The platform provides everything from an easy-to-use content management system to time-stamped commenting and sharing capabilities and
automated highlight reels. Once an Unreel.me site is live, the Bumblebee bot begins including the new content in its analysis and suggestions to viewers. Content owners can choose to monetize their site and create paywalls and subscription plans. Unreel.me also help creators advertise and sell merchandise.
As part of the Unreel.me launch, Unreel Entertainment has raised financing for a USD 1 million Creator Appreciation Fund. The Fund is offering owners of each Unreel.me site 100 percent of all earnings until the money is dispersed. After that, any revenue generated will be split between creators and
Unreel, with content owners retaining 85 percent of their earnings.
Finding ways to reach multiple audiences is a constant challenge, and a number of platforms are helping content creators aggregate the management of their output. We’ve seen an online tool for building apps and widgets for multiple devices and a platform that helps video producers manage content and engage socially all in one place.
What other tools could help content providers manage audience and output?
Tool for build-your-own video streaming sitesUsing AI to find audiences for new content, smart video tech company Unreel Entertainment launches Unreel.me, a platform for custom-built streaming sites.
6
At a Glance
WHATUsing AI to find audiences for new content, smart video tech company Unreel Entertainment launches Unreel.me, a platform
WHOUnreel Entertainment
WHEREUnited States
Powered by
Lawyers have to maintain and recall vast amounts of information in the form of legislation, case law and secondary cases, and they spend up to a fifth of their time on legal research. But an AI app called Ross Intelligence could soon help with that. The program, which is built on IBM's super-computer Watson, uses natural language processing to answer legal questions in a fraction of the time that it would take a legal assistant.
To begin, legal professionals can ask Ross a question as they would ask a colleague. Then the
program reads through the entire body of law and returns a cited answer as well topical readings. Ross also monitors the law constantly to keep the user updated about changes that might affect their case, so they don't need to sift through the mass of legal news.
Ross is already being used by law firm Baker and Hostetler, who will be applying it to their bankruptcy practice. Could a similar program be created for the medical profession?
AI lawyer speeds up legal researchRoss Intelligence, which is built on IBM's super-computer Watson, uses natural language processing to answer legal questions.
7
At a Glance
WHATRoss Intelligence, which is built on IBM's super-computer Watson, uses natural language processing to answer legal questions.
WHORoss Intelligence WHEREUnited States
Powered by
Good design and visual branding is no longer the privilege of companies with the money to pay great in-house designers or outsource the work. We have already seen startups thinking outside the box to provide more affordable design services. This platform offers unlimited, small graphic design tasks to other startups through a subscription package, and Tailor Brands uses artificial intelligence to help business automatically generate a logo that fits their company. Now, Tailor is expanding its tools by launching Tailor Brands Studio — an algorithmic, automated service that
provides companies with unlimited branding packages on a subscription basis.
Tailor Brands Studio will harness data from the 13.5 million designs created by the company's algorithms in the past two years and use it to offer automated designs of logos, business cards, printed stationary and social assets. Companies can sign up for the service on a subscription basis for USD 9.99 a month, which gives them unlimited branding material. The company will add merchandise and ad design later this year.
AI design service can automate a full branding packageAlgorithmic logo-makers Tailor Brands will now generate unlimited branding material for companies via a subscription service.
8
At a Glance
WHATAlgorithmic logo-makers Tailor Brands will now generate unlimited branding material for companies via a subscription
WHOTailor Brands
WHEREUnited States
Powered by
Sifting through reviews to find out how to improve the video games they are building can be a tedious task for developers, but a new program for aggregating game feedback could help. Player XP is a review and analysis platform for mobile gaming.
The aggregator uses machine learning and natural language processing to analyze the reviews. It filters unhelpful reviews or spam, showing only constructive feedback through its artificial intelligence program. As well as removing unhelpful feedback, the platform
categorizes the data via date published or ratings, so more in-depth analysis can be done.
The platform is dedicated to mobile games, and currently aggregates more than 40 million iOS reviews from 250,000 iOS mobile games. Could machine learning be used to help players themselves find new and exciting games that fit their tastes?
Game feedback platform uses machine learning to filter reviewsPlayer XP uses machine learning to help identify constructive feedback in mobile video game reviews.
9
At a Glance
WHATPlayer XP uses machine learning to help identify constructive feedback in mobile video game reviews.
WHOPlayer XP
WHEREUnited Kingdom
Powered by
Being well-informed and able to view things in the longterm can make all the difference when it comes to financial decisions. We have already seen the likes of a tech assistant using big data to help people get the best exchange rates, and now Pefin can help even more. The artificial intelligence platform, which is currently in Beta, is a financial advisor that is significantly cheaper than its human counterparts.
Pefin uses artificial neural networks to constantly monitor over two million data points, and provide data-driven decisions via machine learning. To
begin, customers set up an account and connect it to their other financial accounts, creating a financial profile. They then input their long term plans — such as having children, buying property or going on holiday. These can be changed at any time. Pefin then takes into account inflation, taxes, market conditions, government rules, and even changing daycare costs, and creates an 80 year financial projection. Then, whenever anything changes, either within the customer’s personal finances or in the global economy, Pefin updates the profile and projection automatically.
How else could AI be used to help people plan better for the future?
AI financial advisor makes 80 year projectionPefin is an artificial intelligence financial advisor that monitors over two million data points to provide decisions.
10
At a Glance
WHATPefin is an artificial intelligence financial advisor that monitors over two million data points to provide decisions.
WHOPefin
WHEREUnited States
Powered by
Situated at the intersection of food and technology, NotCo (The Not Company) used artificial intelligence to create the world’s smartest chef. Nicknamed Guiseppe, the advanced algorithm examines the molecular structures of meat and dairy products and attempts to replicate them using plant material.
The company says the secret to creating notably good-tasting vegan meats, eggs, milk and cheeses is to find plants that have similar molecular structures. The final product should have almost identical flavors and textures. When Guiseppe creates a recipe, the algorithm not only considers the relative availability of various plants, it also
tries to enhance nutrition by including additional sustainable ingredients. Products available so far are NotMilk (made from nuts, rice, peas, linseed, and coconut), NotHotdogs, NotChocolate, NotMayo, with more in development.
With animal agriculture using nearly a quarter of the world’s water supply and 45 percent of available land, meat and dairy products are especially unsustainable and environmentally unfriendly. By helping people eat more healthily without compromising flavor and texture, NotCo hopes to lead a revolution in sustainable protein.
So far, much of the work in
food tech has focused either on the processes of buying, selling and delivery, such as paying for coffee with a smart cup, or finding new methods of sustainable production — even if it is in the form of desktop hive farms. NotCo takes the tech directly into the kitchen, creating recipes designed to be used by the average cook.
Are there other industries that could use plant-based, molecular-level replication to increase sustainability?
Artificial intelligence uses algorithms to make nutritious vegan meatChile-based NotCo replicates the taste and texture of animal food products using AI, biochemistry and plants.
11
At a Glance
WHATChile-based NotCo replicates the taste and texture of animal food products using AI, biochemistry and plants.
WHONotCo
WHEREChile
Powered by
In the US, over one million children suffer from autism. These children find it difficult to recognize emotions through facial expressions, making social interactions very challenging. While patients can gain an understanding through behavioral therapy, this can be both time-consuming and expensive. Now, Autism Glass is a wearable aid from researchers at Stanford University, that uses Google Glass, machine learning and real-time social cues to provide those on the autism spectrum with another option.
To use, patients put on the wearable glasses, which incorporate an outward-facing camera. Then, the system uses machine learning and artificial intelligence to understand the facial expressions of people the wearer encounters, and provides cues about their companion's emotions in real-time. The system records the amount of eye contact that takes place, which can provide insights into the wearer's interactions. The wearable is connected to a smartphone app so that the patient, their parents and medical professionals can monitor and
understand their progress and even watch back their interactions.
Research is still undergoing after a successful 40 person pilot study, and the developers are currently seeking participants on the autism spectrum aged 6-16. We have already seen an app that gives users friendship advice by monitoring their heart rate. Are there other new technologies that could help autism patients better understand the world?
AI glasses helps children with autism read facial expressionsAutism Glass is a wearable, behavioral aid that uses machine learning and real-time social cues to help those with autism interact with others.
12
At a Glance
WHATAutism Glass is a wearable, behavioral aid that uses machine learning and real-time social cues to help those with autism interact
WHOStanford University
WHEREUnited States
Powered by
Advertising and media are often at the forefront of new technology, and we have already seen augmented reality platforms showing content in the real world and a virtual reality advertising network for brands.
Now an artificial intelligence robot, AI-CD β, developed by Japanese advertising and marketing agency McCann Japan, is set to work on providing new creative direction for commercials. The AI will give input on projects, mining and analyzing creative databases of adverts to find the best commercials for products
and messages.
The AI-CD β can work through far more historical adverts and commercials than any human advertising creative, and can therefore aid its colleagues in the development of ads for clients. But the robot is also being treated as part of the team at McCann, taking the title of "creative director" and attending the opening ceremony for new company employees.
McCann Japan CEO Yasuyuki Katagi said: “Artificial intelligence is already being used to create a wide variety of
entertainment, including music, movies, and TV drama, so we’re very enthusiastic about the potential of AI-CD β for the future of ad creation. The whole company is 100 percent on board to support the development of our AI employee.”
How else could artificial employees be used to complement, rather than compete, with human co-workers?
In Japan, an artificial intelligence has been appointed creative directorAI-CD β is a robot that is being treated as an actual employee in an ad agency.
13
At a Glance
WHATAI-CD β is a robot that is being treated as an actual employee in an ad agency.
WHOMcCann Japan
WHEREJapan
Powered by
We have already seen headphones that stimulate nerves to let users feel their music. Now, a new speaker system is claiming to use human moods to learn what tunes users want to listen to.
Moodbox uses its artificial intelligence system to work out what music users most want to listen to. Users interact with the Moodbox by voice control, and can keep a diary of their moods using the speaker's smartphone app. The system can work out what music to play by listening to what users say and their tone of voice. Then, depending on the situation
or time of day, Moodbox can make music choices, reacting to, for example, how the user feels first thing in the morning, or when they are coming home from work.
Hong Kong based-developers Ivo say Moodbox uses an emotional intelligence program known as Emi. Emi learns different moods and plays music and lighting accordingly. The program uses AI to analyze the voice of the user and their musical preference, then picks out the perfect tracks from millions of songs and lyrics. The speakers are wireless and
fully voice-controlled, and a number of Moodboxes can be set up in a house and linked together, playing music at the same time.
How else can artificial intelligence be used to improve the way we listen to and discover new music?
Smart speaker reads emotions to play musicMoodbox is an emotionally intelligent speaker system that uses artificial intelligence to learn to pick the best music.
14
At a Glance
WHATMoodbox is an emotionally intelligent speaker system that uses artificial intelligence to learn to pick the best music.
WHOIvo
WHEREHong Kong
Powered by
As search engine rankings become evermore important, companies are hiring costly SEO specialists to help their sites gain more exposure and click-throughs. But for smaller startups that cannot afford to hire someone, Webpage.ly could offer an affordable alternative. The subscription platform aims to help web developers make their sites more searchable.
Webpage.ly uses an algorithm that learns how to boost contextual webpage content over time. It analyzes visitor search behavior and
page rankings to suggest keywords that enable developers to produce higher-impact SEO content. While suggestions are made by the artificial intelligence, users will still have the final say by approving content for their website.
We have already seen AI software being used in the workplace for a variety of functions, from writing personalized emails to hiring the best team fit. How else could artificial intelligence optimize workflow?
SEO artificial intelligence could boost webpage rankingWebpage.ly uses an algorithm that suggests the most searchable keywords, helping developers optimize their sites for better ranking.
15
At a Glance
WHATWebpage.ly uses an algorithm that suggests the most searchable keywords, helping developers optimize their sites for better
WHOWebpage.ly
WHERECanada
Powered by
Finding new video content online, or searching for films without knowing the title can still be a challenge, even with the most up-to-date search engines. That's why Val.ai, a new video search engine from Finnish company Valossa, is using an artificial intelligence-powered tool that can analyze thousands of variables in movies and videos, and give more accurate searches.
Val.ai is capable of analyzing video streams and identifying thousands of themes — the artificial intelligence looks deep into the content, with the
company claiming their software as "the most advanced video analysis platform".
The platform can examine advanced questions and queries to find films. Users can type in, for example, "show me sci-fi movies about space battles and laser guns", or "find me epic history movies". The platform also features a descriptive voice-controlled search engine for discovering entertainment platforms.
How else can search engines be improved to make smarter and more useful searches?
Filmbuff search engine helps users find movies they likeVal.ai is an artificially intelligent search engine for films, which can process natural language and analyze thousands of variables.
16
At a Glance
WHATVal.ai is an artificially intelligent search engine for films, which can process natural language and analyze thousands of variables.
WHOValossa
WHEREFinland
Powered by
Moderating social media feeds or online comments is important for marketers and entrepreneurs, but looking after a business’s social media accounts can be incredibly time consuming. Now Smart Moderation is bringing intelligent automated moderation to Facebook and Instagram profiles.
The program uses an AI software that understands natural language. It processes the meaning of sentences, rather than just flags for keywords, and removes trolling or abusive comments accordingly. The company says the software learns and
reviews like a human. The dashboard lets users view all comments from different social pages, and can be set to automatically hide inappropriate comments. Users have the option to receive daily reports of spam comments, and can also switch to manual moderation. The platform performs more traditional spam-blocking tasks too, such as clearing adverts.
We have seen artificial intelligence being used to turn spreadsheets into written reports and clean up inboxes. How else can smart online programs streamline the process of managing businesses'
digital profiles?
Artificial intelligence learns to block spam messages and commentsSmart Moderation uses artificial intelligence to block spam messages and comments on social media.
17
At a Glance
WHATSmart Moderation uses artificial intelligence to block spam messages and comments on social media.
WHOSmart Moderation
WHEREUnited States
CONTACTwww.smartmoderation.com [email protected]
Powered by
We’ve seen how algorithms are being used to help build strong startup teams, and now Test4startup can help founders test the 'successability' of their actual idea using artificial intelligence software.
Users begin by uploading an overview of their business aims. Test4startups then implements a keyword-based analysis that aggregates information relative to the specific market, providing infographics of competition, trends and pricing variation. Test4startups algorithm is then able to provide
recommendations based on how a product or service compares to others — recommending a price reduction, for example. Startups can test their products and business ideas from USD 10, or for USD 15, they can also receive feedback from focus group professionals.
How else could algorithms be used to analyse context and provide feedback?
AI algorithm predicts startup successTest4startup uses an algorithm to test business idea strength in current markets and against competitors, generating feedback and advice.
18
At a Glance
WHATTest4startup uses an algorithm to test business idea strength in current markets and against competitors, generating feedback
WHOTest4startup
WHERERussia
Powered by
Every year in Switzerland, emergency centers respond to around 1,000 call outs for lost and injured hikers. It can often take hours and significant manpower to locate lost mountaineers, but new software for quadcopter drones is making the hunt quicker and easier, and has the potential to help find human survivors in disaster zones around the world.
The drone uses a computer algorithm called a Deep Neural Network. The program was developed by researchers at the University of Zurich and the Dalle Molle Institute for Artificial Intelligence. The drone
uses the algorithm to learn trails and paths through a pair of small cameras, interpreting the images and recognizing man-made pathways. Even when working on a previously unseen trail, it was able to guess the correct direction in 85 percent of the cases. The drones' speed and accuracy make them more effective than human trackers.
The researchers hope that eventually multiple small drones could be combined with human search and rescue missions, to cover more terrain and find people faster. The drones can cover terrain quickly and check hazardous
areas to minimize risk to human workers, and its AI can identify paths and avoid crashing without any human involvement.
However, the developers say the technology is still in its early stages, with new software R&D required to make the drones able to identify human targets and relay this back to rescue teams.
We've already seen ambulance drones helping citizens give lifesaving procedures before professionals arrive. How else can drones and robots help rescue workers save more lives?
Drones better than human rescuers at following mountain pathwaysThe Deep Neural Network software helps quadcopter drones identify trails and paths, and lead them to lost or injured hikers.
19
At a Glance
WHATThe Deep Neural Network software helps quadcopter drones identify trails and paths, and lead them to lost or injured hikers.
WHOUniversity of Zurich
WHERESwitzerland
Powered by
Instant messaging app Telegram is growing in popularity with its multi-platform integration and encrypt-able P2P messaging. Telegram programming has an open API so users can add bots into chats, and one such feature is MurdochBot, which hopes to suggest TV shows users will like.
Murdochbot is an AI protocol that uses sentiment-analyzing algorithms developed by IBM, in combination with Sky TV’s archive of viewer data. Users first add MurdochBot into any conversation, and it will ask them a few basic
questions about their favorite TV shows. The bot then monitors the content to understand not only what kind of shows users will like, but also in what context and times they would like to hear about them, learning more as conversations develop. Founder Ben Dixon believes that analyzing natural conversations will give businesses a better idea as to what consumers want. Unveiled at the TechCrunch Disrupt London Hackathon this year, Murdochbot has yet to reveal development plans.
As instant messaging
eclipse social media by offering more personal connectedness, a new space is created for data-mining and delivering content. How could companies make the most of this opportunity, while providing users with autonomy over their privacy?
Bot suggests TV shows based on group chatsMurdochBot uses IBM's sentiment algorithm, Sky TV viewer data and users' conversations to offer entertainment recommendations.
20
At a Glance
WHATMurdochBot uses IBM's sentiment algorithm, Sky TV viewer data and users' conversations to offer entertainment recommendations.
WHOMurdochBot
WHEREUnited Kingdom
Powered by
There has been much talk about driverless cars, and we're seeing an abundance of assistive driving technology, such as this software for car-to-car communication. To demonstrate just what driverless cars are capable of, electric vehicle racing series Formula E have partnered with Kinetik to announce the world’s first driverless car race.
The Roborace, scheduled for the 2016/17 season, will involve ten teams using the same basic driverless vehicles — the challenge will be in developing the competitive software. Teams will develop artificial
intelligence that must be capable of dealing with real-time obstacles, overtaking maneuvers, and weighing up risk and reward scenarios. The races will take place on Formula E circuits in major cities around the world, with the aim of showcasing the performance and ability of driverless technology in the extremes of speed and risk. Besides the professional teams, at least one team is to be crowdsourced.
Further details are set to be announced in early 2016. First person drone racing has already been around for over a year.
How else can autonomous robotics be applied to sports?
Formula E announce first driverless car championshipKinetic, in partnership with Formula E, are developing driverless cars that will use AI to compete in races from 2016.
21
At a Glance
WHATKinetic, in partnership with Formula E, are developing driverless cars that will use AI to compete in races from 2016.
WHORoborace
WHEREFrance / United Kingdom
Powered by
A new offering from Automated Insights will analyze data and write personalized reports.
The service, called Wordsmith, is currently in private beta and already being used by giants like Yahoo!, Comcast, Samsung and others. To begin, users first upload a spreadsheet or an alternative form of numerical data. The system will then convert the individual fields into variables, which are used populate a text template. Users can also apply rules to the data to create certain phrases or sentences within the text. For example, if a
number is greater than a certain value, a specific set of words can be used. Namely, if the sales from this month’s datasheet were greater than last month’s, words like “It was another great month” can be inserted. The users will still have to know which data correlates to which information, but once the various rules are preset, the task will become much more streamlined.
We have already seen an AI-powered tool that can turn legal contracts into simple visual charts, and a service that will share a post on multiple platforms with one action. What other
administrative tasks can be automated through similar innovations?
Software turns spreadsheets into written reportsAutomated Insights have developed a free system that can analyze data to write personalized reports.
22
At a Glance
WHATAutomated Insights have developed a free system that can analyze data to write personalized reports.
WHOAutomated Insights
WHEREUnited States
Powered by
The future of wearables is becoming increasingly smart and personal, with devices able to recognize and track multiple exercises and offer useful feedback. Moov Now takes these abilities to the next level: the wearable fitness tracker and AI personal trainer can help users train for five different sports, by measuring their movements within 3D space.
Moov Now is the next generation of Moov's hugely successful fitness wearable, which can be used to enhance training for running, swimming, cycling, boxing and a seven minute bodyweight
workout. To begin, users sync the wristband with their smartphone or tablet. They select their chosen exercise and the device monitors their first few work outs, then places them on a level: running for example has 69 different levels depending on fitness and goals. The wearable then uses a 9-axis Omni Motion™ sensor — combining an accelerometer, a gyroscope and a magnetometer — to capture the wearer's motions in 3D. An algorithm will then processes the data, allowing the band to measure speed, stride cadence, elevation,
velocity, impact and more depending on the activity.
Moov Now, like its predecessor, has an inbuilt coach, who offers real time feedback by talking to the user during their workout, motivating them and safely pushing them towards goals. Users who prefer to workout to music can integrate the app with Spotify so that they can enjoy both.
Could these capabilities be used for professional sports or dance, too?
Multi-sport wearable coach offers intuitive motivationMoov Now is a wearable fitness tracker and AI personal trainer, which identifies exercises by measuring movement within 3D space.
23
At a Glance
WHATMoov Now is a wearable fitness tracker and AI personal trainer, which identifies exercises by measuring movement within 3D
WHOMoov
WHEREUnited States
CONTACTwww.welcome.moov.ccwww.twitter.com/GetMoov
Powered by
We have seen a host of work related apps that aim to make tedious office tasks more approachable — there is a plugin that can find files without knowing the title, and a tracking tool which analyzes competitors online strategies. Joining this is Beagle, an intelligent contract analysis tool which provides users with a graphical summary of lengthy documents in seconds. It is a time-saving tool which translates complicated documents from elusive legal language into comprehensive visual summaries.
The Beagle system is powered by self-learning artificial intelligence which learns the client's preferences and adapts accordingly. Users begin by dropping in a file into the app. The AI — trained by lawyers and NLP experts — then converts the information into a single page document. It processes the contract at a rate of one page per 0.05 seconds and highlights key information, displaying it in easy to read graphs and charts. The system also comes with built-in collaboration tools so multiple users can edit and export the files.
What other time-consuming work tasks could be made more efficient by artificial intelligence systems?
AI tool turns complicated legal contracts into simple visual chartsBeagle is an AI-powered tool that translates contracts from elusive legal language into comprehensive visual summaries.
24
At a Glance
WHATBeagle is an AI-powered tool that translates contracts from elusive legal language into comprehensive visual summaries.
WHOBeagle WHERECanada
Powered by
Most people interact with their smartphones in recurrent routines — consistently launching their weather app first thing in the morning, or always booking an Uber after checking Citymapper late at night. Snips is an intelligent interface for smartphones which will learn these regular behaviors and perform them automatically, enabling the user to access the information and services they need without having to search for them.
Snips employs big data and AI to monitor and analyze the smartphone habits of users — learning
their patterns and enabling faster, more ubiquitous access to services. Their aim is to make the technology work for consumers without the need for conscious interaction: to make technology more context-aware and therefore much less visible. We have seen other platforms — such as IFTTT — which enable customers to automate online tasks via creative 'recipes', but Snips takes this to the next level by automating the whole process.
Snips is currently in Beta and is welcoming participants. How else
could Snips' AI be used to improve consumer's experiences?
Context-aware interface will automatically launch the right app at the right timeSnips uses big data and AI to learn users' smartphone habits and automate their interactions with their devices.
25
At a Glance
WHATSnips uses big data and AI to learn users' smartphone habits and automate their interactions with their devices.
WHOSnips
WHEREFrance
Powered by
We have seen a number of initiatives — such as Supply Change — determined to curb deforestation by putting pressure on businesses to adjust their supply chains. Taking another tack, startup Orbital Insight are using big data and satellite images to monitor forests and discover threatened areas before any illegal deforestation is allowed to take place.
Orbital Insight have collaborated with Global Forest Watch on a system which monitors and recognizes suspicious changes, such as extensive new road building, around
forests. Once a threatened area has been identified, alerts are published on the Global Forest Watch website, which local people and other conservation organizations use as a resource. The hope is that, by flagging up worrying developments in good time, the system can help to reduce the huge areas of forest destroyed illegally every year.
The AI system is using deep learning to develop its understanding of warning signs — meaning that the longer it monitors the globe, the more patterns the neural network will recognize, and
the better it will become at its job. Orbital Insight also use their technology to help corporations predict global trends. How else could companies like Orbital Insight use their satellite images for good?
AI startup spots deforestation from the sky — in time to stop itOrbital Insight are using big data and satellite images to monitor forests and discover threatened areas before illegal logging takes place.
26
At a Glance
WHATOrbital Insight are using big data and satellite images to monitor forests and discover threatened areas before illegal logging takes
WHOOrbital Insight
WHEREUnited States
Powered by
Most contact lists are shockingly out of date — an average of 25 percent of numbers and email addresses stored in people's digital contact books are no longer in use. In business, this can be a very expensive mistake. Helping to solve this, CircleBack is a simple app which uses artificial intelligence and crowdsourced data to enable users to keep their business and personal contacts up to date.
CircleBack is a simple platform for individuals and businesses that manages users' address books, automatically deleting duplicate contacts, filling in missing details and updating
records when a contact's information changes. To begin, individual users download the app or sign in online. The platform then connects with a range of social networks, including Facebook and LinkedIn, to collate the user's contacts and creates a single, secure, cloud-based address book which they can access anytime from any device. Users can also sync CircleBack with their Gmail account using ContactSaver, converting email signatures into usable information.
The platform's artificial intelligence engine then continuously processes billions of data points to determine whether contact information is accurate
and up to date. CircleBack already has a database of over half a billion members from the global business community, but since the information comes from the user base itself, the more people who use CircleBack the more accurate it will become. CircleBack processes contacts anonymously to ensure privacy is protected.
The platform is free for individual users and available to businesses for a fee. Are there other services which could help people to spring clean their online data?
AI contact manager keeps users' digital address books up to dateCircleBack manages users' address books, filling in missing details and updating records when their contacts' information changes.
27
At a Glance
WHATCircleBack manages users' address books, filling in missing details and updating records when their contacts' information
WHOCircleBack
WHEREUnited States
Powered by
Research in AI tends to focus on developing robots who are more capable and efficient than humans, but a new study at École Polytechnique Federale de Lausanne in Switzerland is doing just the opposite. When children teach their peers they gain motivation and self esteem — and ultimately they improve themselves, all without realizing. The CoWriter Project has created a robot classroom assistant who is less advanced at handwriting than the most struggling students — giving those pupils someone to learn with and teach.
Scientists from the
Computer-Human Interaction Lab at EPFL used the like-able humanoid NAO robot in their research with school children aged six to eight. NAO can be programmed to reproduce common handwriting errors, which it learns from a database of examples of children's writing. To begin, the child creates a word from small magnetic letters and the robot writes out the word on a tablet screen. The child then identifies any errors the robot has made and corrects them, teaching the robot how to write better and improving their own writing in the process. The robot repeats the task, using algorithms
which lead to gradual improvement. Once the child is satisfied, they move on to another word. The robot can even be made to adopt a child's specific writing difficulties — such as the letter F or H.
The CoWriter system is currently in the prototype stage, but it has been very well received at a primary school where it has been used in lessons with about 70 students. The developers plan to continue testing its effectiveness and benefits in the coming months. Are there other applications for a robot which enables children to teach the skills they are struggling with?
Teaching this robot will help children to improve their handwritingIn the CoWriter project, a humanoid robot is the least advanced pupil in class — giving struggling students someone to learn with and teach.
28
At a Glance
WHATIn the CoWriter project, a humanoid robot is the least advanced pupil in class — giving struggling students someone to
WHOComputer-Human Interaction Lab at EPFL
WHERESwitzerland
Powered by
Last year we wrote about Jibo — the world's first family Robot — a friendly AI device which uses facial recognition and natural language processing to offer personal assistance in the home. Jibo is a table top device with more than a passing resemblance to Pixar's endearing animated lamp, that is programmed to become a social and interactive member of the family — taking photographs, connecting with other smart devices in the home, and talking to its owners in a Siri-like voice. Now, Robotbase are crowdfunding their own take on the friendly artificial intelligence
assistant in the form of the Personal Robot.
The Personal Robot shares a vast number of features with Jibo — its facial and object recognition capabilities, its ability to correspond with users intelligently, its potential interaction with the Internet of Things and its capacity to provide entertainment — changing lighting, playing music and narrating stories to younger users. The main advancements on previous devices are its ability to move around by itself and its ability to detect its owner's emotions. Its mobility and intelligent navigation are an important
evolution, enabling it a much greater degree of autonomy and allowing it to move from one room to another as and when it is needed. It uses mapping and navigation algorithms to build its own map of the house or office and there-after can also provide real-time video surveillance — adding personal security guard to its long list of occupations.
Robotbase are also keen to highlight the devices potential capacity as an office assistant — it can access facts and data and take meeting notes — and are offering customized Personal Robots with
Autonomous robot is personal assistant, cameraman and home security system in onePersonal Robot, the latest humanoid AI device from Robotbase, adds autonomous navigation and emotion recognition to its long list of helpful capabilities.
29
At a Glance
WHATPersonal Robot, the latest humanoid AI device from Robotbase, adds autonomous navigation and emotion
WHORobotbase
WHEREUnited States
Powered by
company logo and uniform to companies who want to road test the office of the future.
Personal Robot is a 4ft upright model. Its digital 'face' displays the head and shoulders of one of numerous friendly animated cartoons. The 'face' is connected to the mobile base by a simple white pole so that the whole device resembles an oversized board game piece. As more humanoids appear on the marketplace, appearance and "personality" will doubtless become deciding factors in consumer selection. What other forms could the personal robot take to appeal to different customers?
30
At a Glance
WHATPersonal Robot, the latest humanoid AI device from Robotbase, adds autonomous navigation and emotion
WHORobotbase
WHEREUnited States
Powered by
Dealing with the customer support department of many companies can be an infuriating task -- as a recent viral recording of a Comcast call demonstrated pretty perfectly. But can digital alternatives offer an improvement? IPsoft’s Amelia is an AI customer service agent that can accurately detect the mood of the customer to better deal with their request.
While many businesses have already experimented with artificial intelligence for online web support, it can soon become very obvious when the 'agent' doesn't respond in an appropriately emotional
way, especially if the request is an angry one. Amelia is able to process the natural language of both business manuals and the customer in need of support. Not only can it search those business guides and the web for information to help each customer, but it also tracks their mood and other contextual data. This way, if a customer is unhappy, Amelia tailors its support to offer greater reassurance and works to ensure customers end the call in a better mood. If in any case the issue isn't resolved, Amelia passes the call or chat onto a human. All communication is stored
so Amelia can pick up each case with reference to any previous contact.
Detecting customers’ feelings and responding appropriately could make online helpdesks even more like face-to-face support, while helping reduce the burden of customer service teams on businesses. Are there other ways this type of mood-detecting technology could help digitize human responsibilities?
Virtual customer service uses AI to detect human emotionIPsoft’s Amelia is an AI customer service agent that can accurately detect the mood of the customer to better deal with their request.
31
At a Glance
WHATIPsoft’s Amelia is an AI customer service agent that can accurately detect the mood of the customer to better deal with their request.
WHOIPsoft
WHEREUnited States
CONTACTwww.ipsoft.comwww.ipsoft.com/contact
Powered by
Finding a new place to eat with a friend or that special someone often requires a trawl through Google search results or Yelp reviews, but they're not very good at knowing the types of restaurants you already like. Adding a touch of natural language recognition and artificial intelligence, Russia's IO is an app that learns users' preferences and chats with them to help them find restaurants nearby.
Users simply start by typing questions into IO, without having to fill out any preferences up front. IO will then begin to reply with an easy-to-understand answer in a friendly tone. Users can type as if they were speaking
to a friend over text and IO will understand what they're saying. When recommending a restaurant, the app brings up all of the relevent details -- location, description, menu, prices, opening times and contact details. Users can also ask what's good on the menu, what the atmosphere's like and if it's good for a date, along with other questions they might have. For each response, users can let IO know if they think it looks good, if they want to know more info or a different suggestion. This way, the app learns their preferences and can make better recommendations in the future. The more they use it, the smarter it becomes. Over time, users can save
information about their favorite restaurants to create their own unique collection.
Recently launched in New York City and with plans to expand to major European cities, IO is part of a trend for presenting data in a more friendly way using AI and machine learning, and at times it feels close to the vision of the future presented in Spike Jonze's Her. Are there other ways to make tech user interfaces feel less like working with a computer and more like speaking with a human?
AI restaurant guide makes friendly recommendationsUsing natural language recognition and artificial intelligence, Russia's IO is an app that learns users' preferences and chats with them to help them find restaurants nearby.
32
At a Glance
WHATUsing natural language recognition and artificial intelligence, Russia's IO is an app that learns users' preferences and
WHOIO
WHERERussia
Powered by
Not everyone has an eye for good design, or the skills to pull it off. And yet it’s an incredibly important aspect of any business, especially in the age of the web where judgments are made at lightning speed. Blogging platforms such as Wordpress and Tumblr, along with WYSIWYG, do-it-yourself builders, have made website creation much easier in the last few years, but they still require a ton of human input and control. We recently wrote about Tailor, which uses AI to create contemporary branding, The Grid is a site-building platform that intelligently analyzes content and structures it
in a beautiful and easy-to-read layout.
The service enables customers to choose a ‘layout filter’ — essentially a theme — for their site, but rather than content slotting into that design, the design shifts dependent on the content. For example, the system analyzes images for color and alters the color of the caption box to a shade that’s complementary to both the image and the site theme. Images can also be cropped to perfectly frame the content through facial and object recognition, as well as automatically resized and programmed
for responsive design.
The Grid also asks users to define a purpose for their site — whether it’s to gain more followers, sell a product or display a portfolio, among others. The layout is then adapted for each objective. Each theme can automatically pull in related links and content from the web depending on keywords in post text. The layout also shifts dynamically to emphasize new content, as well as posts that are being shared on social networks or are linked to trending stories.
According to the creators,
AI websites design themselvesThe Grid is a site-building platform that intelligently analyzes content and structures it in a beautiful and easy-to-read layout.
33
At a Glance
WHATThe Grid is a site-building platform that intelligently analyzes content and structures it in a beautiful and easy-to-read layout.
WHOThe Grid
WHEREUnited States
Powered by
a Wordpress site takes on average 18 hours to build and customize. The Grid can create an equivalent site in 3 minutes, although users need to be happy to surrender some creative control over to a robot. Although the concept uses artificial intelligence — and in some cases machine learning — to tailor sites to each user, there’s always a real designer on hand to ensure layouts don’t go rogue and create a design that looks awkward to human eyes. For those who want to try it out, the service will cost USD 25 a month, which is good for creating up to 7 sites. Are there other ways that AI could make content production and presentation an easier and less time-consuming task?
34
At a Glance
WHATThe Grid is a site-building platform that intelligently analyzes content and structures it in a beautiful and easy-to-read layout.
WHOThe Grid
WHEREUnited States
Powered by
While general elections in the UK typically receive a moderate voter turnout — usually above 60 percent — there’s still widespread disaffection with politics and voter numbers have been dropping since the 1950s. Young people especially aren’t engaged by political activity in their region and learning about it isn’t mandatory in schools. Now a new campaign in the country is looking to change this with Ask Amy, an app that uses artificial intelligence to provide kids with answers to their questions about politics.
Designed by the Tell Me
About Politics campaign, Ask Amy is designed to look just like a messaging app such as WhatsApp, and users can send any message they like at any time of the day. Amy — who is powered by natural language processing and artificial intelligence — is able to make sense of the question or concern and deliver relevant and unbiased information in a friendly way. Rather than plain text messages, Amy is also capable of pulling up images, videos and even interactive content. For example, a young person could tell Amy that they don’t see the difference between each
of the parties they can vote for, and they’ll receive a prompt to take a Buzzfeed-style quiz to determine which one best represents their views.
Ask Amy has already been trialled among young voters. While the AI perhaps gives attention-bereft youngsters instant replies to their questions, could this kind of app benefit from using real people to provide more tailored advice to potential voters?
Through app, kids ask questions about politics via textLooking to engage young people with politics, Ask Amy is an app that uses artificial intelligence to provide kids with answers to their questions about the subject.
35
At a Glance
WHATLooking to engage young people with politics, Ask Amy is an app that uses artificial intelligence to provide kids with answers to their
WHOTell Me About Politics campaign
WHEREUnited Kingdom
CONTACTwww.askamy.netwww.tellmeaboutpolitics.com/
Powered by
To-do apps are a popular way to ensure that our increasingly busy lives stay organized, but they can become so sprawlingly as to actually take up time themselves. We recently wrote about Timeful, which uses smart suggestions to help users plan their day more effectively, and now x.ai is a service that enables busy people to task an AI personal assistant to deal with the tedious back and forth involved in organizing a meeting.
Many people will be familiar with the difficulty of arranging a time and place that's good for both
or all parties wanting to have a get-together. With x.ai, anyone can hand over the boring task of negotiating the details when a colleague requests a meeting. Users simply reply to the original message while CCing in the email address of 'Amy' — [email protected] — making sure to include a Siri-style request such as 'Amy, please find a time for me to meet with [colleague] on Wednesday'. Amy uses information from the user's calendar to find a time slot and location that matches the request and creates an email in natural language as a proposition to the colleague.
Amy also takes into account preset preferences such as Skype and Google Hangout availability and favored locations when negotiating with other parties. For example, if the colleague asks to meet at their office, Amy will politely ask if there's a chance it could take place at the user's office instead. The emails continue back and forth until an agreement is met. Amy then sends a meeting invitation to both parties that they can add to their calendar.
x.ai is capable of arranging meetings between up to 5 people, and also works
AI robot called Amy will politely arrange meeting requests with your co-workersx.ai is a service that enables busy people to task an AI personal assistant to deal with the tedious back and forth involved in organizing a meeting.
36
At a Glance
WHATx.ai is a service that enables busy people to task an AI personal assistant to deal with the tedious back and forth involved in
WHOx.ai
WHEREUnited States
CONTACTwww.x.aiwww.x.ai/contact
Powered by
if the other party uses the service. In this case, there isn't even any dialogue and the arrangement is automatically worked out according to the calendars of each user. Multiple calendars can be added — both business and personal — to ensure meetings don't eat into leisure time. Customers can even change the name of the assistant if they don't like Amy.
x.ai requires no installation -- simply email the address to get set up -- and is currently free to use, although the team say a subscription fee may be added at some point in the future. Are there other ways to leverage AI to do some of the more menial human tasks?
37
At a Glance
WHATx.ai is a service that enables busy people to task an AI personal assistant to deal with the tedious back and forth involved in
WHOx.ai
WHEREUnited States
CONTACTwww.x.aiwww.x.ai/contact
Powered by
The death of a loved one is an irreparable loss, but tokens of the departed can help grievers to remember what it was like when they were alive and appreciate their existence. While projects such as DNA Memorial enable future generations to learn more about their passed relatives' genetic makeup, a new startup called Eterni.me is going much further. By trawling traces of their online activity, the company wants to reconstruct people's personalities with an artificially intelligent version that friends and family can 'chat' to after they're gone.The uncanny idea — which bears a striking resemblance to an episode of the UK's Black Mirror science fiction
series, which explores the dark implications of digital technology — collects together chat logs, text messages, emails and social media posts from the deceased and uses algorithms and artificial intelligence to present what is essentially a chatbot that resembles that person's online persona. Although the project — which has come out of the MIT Entrepreneurship Development Program — is still in its early stages, users will be able to speak or type questions and prompts to an avatar of their dead relative, who will respond with information gleaned from their online past. According to its creators, the concept is less about ressurrecting the dead and more about
unlocking the memories hidden in the trails of their internet usage. If that sounds creepy, Eterni.me CEO Marius Ursache recently assured Fast Company: "We're very aware we're not creating a digital clone or anything creepy."
Eterni.me's essential aim is to lessen the impact of death and ensure that people are remembered after they die — something everyone hopes for. It also taps into a wider trend for dealing with the masses of social media data that's left after we're gone. Are there other ways to create tributes to friends and family members using the memories they leave behind?
AI resurrector lets people Skype their dead relativesBy trawling traces of their online activity, Eterni.me wants to reconstruct people's personalities with an artificially intelligent version that friends and family can 'chat' to after they're gone.
38
At a Glance
WHATBy trawling traces of their online activity, Eterni.me wants to reconstruct people's personalities with an artificially intelligent
WHOMarius Ursache (MIT Entrepre-neurship Development Program)
WHEREUnited States
Powered by
Calendar management has become a perennial challenge for many people in today's time-strapped world, but the basic model for such apps – in most cases, the simple paper calendar – has changed little over the years. Enter Tempo, a smart calendar app from SRI International that uses artificial intelligence to make users' schedules more manageable.
Launched by California-based Tempo AI, a spinoff of SRI, the Tempo Smart Calendar is a mobile productivity app that analyzes data from a user’s mobile calendar and applies artificial intelligence and machine learning to enhance meetings and events throughout the day.
Specifically, it's designed to find and neatly organize in context everything the user needs to be fully prepared, including contacts, relevant emails, and related documents. Tempo can handle the passcodes for dialing into conference calls, for example; it can also open Word, Excel, PowerPoint and PDF documents in preparation, as well as browsing attendees' LinkedIn profiles and scouting locations with Foursquare and Yelp. In addition, Tempo can find meeting locations along with driving directions, nearby parking, and estimated drive time. Users can even send a preset email or text if they find themselves running late. Perhaps best of all is that the more Tempo
is used, the more fine-tuned and personalized it becomes because it learns and recognizes patterns to make better suggestions over time.
The company explains: “Using Tempo Smart Calendar is like having a personal assistant prepare for what’s next – not only providing the right information, but anticipating and understanding what a user intends to do with it. With details in the event summary, it is simple to quickly find information, eliminating the issue of hunting around other apps, email or the browser while trying to complete a simple task such as finding a meeting agenda or someone’s phone number.”
Personal calendars given an artificial intelligence makeover through iPhone appTempo is a smart calendar app from SRI International that uses artificial intelligence to make users' schedules more manageable.
39
At a Glance
WHATTempo is a smart calendar app from SRI International that uses artificial intelligence to make users' schedules more
WHOTempo AI
WHEREUnited States
Springwise1 Sekforde StreetEC1R 0BE, LondonUnited Kingdom
Copyright © 2016 Springwise Intelligence Ltd. All rights reserved.
Powered by