Artificial Intelligence: The force awakens
First things first: what is AI? If you look around, there are tons of definitions but the one given by Goldman Sachs in their equity report seems clear and complete:
“AI is the science and engineering of making intelligent machines and computer programs capable of learning and problem solving in ways that normally require human intelligence. Classically, these include natural language processing and translation, visual perception and pattern recognition, and decision making, but the number and complexity of applications is rapidly expanding”
AI is trendy nowadays but not new. Back in 1997, Deep Blue, a chess-playing computer developed by IBM defeated Garry Kasparov who is still considered as one of the greatest chess players in the world.
But what has changed since? The proliferation of data and the rapidity of computing power have accelerated the development of algorithms. Machine learning emerged as a branch of Artificial Intelligence and allows computers to learn from datasets without being programmed explicitly. Deep learning is defined as a subset of machine learning and enables systems to solve more complex problems. Deep learning involves training multiple hidden layers on how to resolve a part of the case.
Where can we see AI more concretely? Almost everywhere and this is the theme of this post. Every application now tends to have a smart layer in order to analyse and predict behaviours. At work or at home, AI will soon become the norm to some extent.
In order to embed some AI in an application, it is almost necessary for the machine to understand our language, this technology is known as NLU or Natural Language Understanding. This implies for the computer, among other things, to get the structure of the sentence (syntax) and to capture the meaning (semantics), with the latter being the most complex. This requires that the machine digests a large number of datasets and rules.
The GAFAs saw quickly the opportunity and developed their NLU platforms, often by acquiring emerging startups in the field. For instance, Facebook snapped Wit.ai up in January 2015 and Google bought more recently API.ai (September 2016). Independent players try to get their slice of the cake too such as Recast.ai in France or RASA in the UK which is open-source. In the meantime, Microsoft is building its own tool called LUIS.ai (part of its Cognitive Services platform) and IBM acquired AlchemyAPI in March 2015 to strengthen its Watson branch.
NLU technology is paramount for chatbots and virtual personal assistants where the user now interacts with a messaging or voice platform. Applications are infinite since instead of performing several actions on different programs, the user will only have to engage with a single interface. Bots are popping up everywhere on websites to facilitate customer support or e-commerce searches. They are even present on Facebook Messenger (check out Citron if you are looking for a trendy restaurant tonight!) and will be soon released on WhatsApp.
The natural language output produced by the bot (step 5) comes from the Natural Language Generation (NLG) technology, which turns structured data (step 3) into a qualitative text. NLG use cases go beyond simple bot answers: they can help write newspaper articles or product information sheets (have a go on LabSense website!).
Technology giants now want to exploit Artificial Intelligence algorithms to improve their conversational assistants and integrate them in their ecosystem. For instance, Microsoft developed Cortana to equip its Windows phones, Xbox and so on. Apple is continuously enhancing its proprietary vocal assistant Siri; they bought two years ago the young UK-based company Vocal IQ and are now opening Siri to external developers. Google deployed its Google Assistant on its Pixel phones and Nest products and is now releasing it to Android phones. As for Samsung, they completed the acquisition of Viv Labs in February 2017 for approximately $210m.
But the most advanced company in the field is probably Amazon which developed its Echo device, powered by its personal assistant Alexa. Echo can currently be ordered in the US, UK and Germany and assists its owner to order food or a cab, play music and even announce weather forecasts or sport scores.
Snips, a French startup, took the opposite view to what Amazon did with Alexa. They allow users to interact with personal assistants with text or voice with the data staying locally on the device and hence protecting their privacy.
Some AI companies can analyse text and in parallel, pictures. This is the challenge that Heuritech tackles with its solution for luxury groups. Their algorithm is able to identify trends around fashion items and celebrities in both text and image.
Some startups preferred to tackle the computer vision market. This is the case of Angus.ai which assists retailers to understand what is happening in their stores and how customers behave in front of the shelves and special offers. Smart me Up exploits another use case and plugs its solution in outdoor cameras to help smart cities being safer. Computer vision technology may also be applied for the development of the self driving car. Tech giants such as Google, Uber and even Apple are working on autonomous cars, which will literally revolutionise the way we commute and live. Uber bought Otto last August for $680m and is now accused by Google to have stolen its IP.
AI is expected to irrigate all the other industries (retail, fintech, medtech to name a few) and tech giants will continue buy and build to develop their AI solutions. Nevertheless, some enterprises will have to rely on AI services providers as they will not have enough deep tech experts and will lack clean datasets to train a model. AI-aaS is an emerging and interesting concept and some startups have taken the lead in this field. Craft ai for instance offers APIs for connected objects manufacturers and mobile applications providers.
We strongly believe that the tech scene is buoyant in France and that is only a beginning. AI is increasingly present in our daily routines and early-stage companies in this domain are flourishing in the country. World-class engineering schools, deep tech talents and R&D tax credits give France an edge to become an influent AI centre. Companies such as Cisco and Facebook saw the ecosystem maturing and recently set up AI operations in Paris.
At Apparius, we assist startups on a day-to-day basis, help them find the right use cases for their technology and craft their funding strategy. We are always happy to chat about AI, tech or anything else over a coffee, so feel free to ping us!
