1 Introduction This study set out to identify Radical Innovation Breakthroughs ~ potentially important cross-cutting, disruptive innovations - that could be usefully considered in the development and implementation of: © European Union Research and Innovation policy, especially in the implementation of Horizon Europe, © European Union policies that are supported by the framework programme (e.g. security, environment, agriculture, health, etc. - see Article 179 TFEU) and © relevant strategies at national and regional level, including smart specialization. Radical Innovation Breakthroughs are not instant. They result from cumulative processes of multiple inventions and innovations, and their diffusion and adaptation to different circumstances. Their key characteristics are (a) an unusually high impact potential and (b) the potential to disrupt technical, economic and social structures. This study searched for future Radical Innovation Breakthroughs in literature associated with the continuous efforts of people and society to innovate in technology such as peer reviewed journals from a range of S&T disciplines and platforms sharing cutting-edge S&T news. In this sense, most of the RIBS identified are technological innovations. People often draw a distinction between technological and social innovations. New movements in society create knowledge about how to do things differently, often emphasizing behavioural aspects rather than the role of technical artefacts. This does not mean that technology does not play a role in social innovations. The continuous effort for social innovation and improvernent is part and parcel of human societies, but it is not organized in the same way as the effort for technological innovation. When searching for social innovations in literature, we came across trends that are already well diffused and established. However, some of them contradict fundamental elements of social norms and systems and have thus considerable disruption potential, whilst opening up possibilities to solve important problems plaguing contemporary societies. The way in which technological and social innovations will combine to shape future industries and sociotechnical regimes can be a matter of considerable uncertainty and speculation. We see so much disruptive potential around us that we are confident that the industries, sectors and value-chains of the future will not be like the ones of today. In order to evaluate the potential impact of future innovations, we considered a number of future global value creating structures, which we termed Global Value Networks, following the approach of the Finnish Radical Technology Inquirer (RTI).! These structures represent a mixture of expectations (how things will be), wishes (how things should be) and appreciations of how things are. In each case, the defining characteristic is a network that is associated with a function for which there * Lintur, Risto, Osmo Kuusl and Toni Ahlqvist (2014) 100 Opportunities for Finland and the World, the Radical Technology Inquirer (RT), the English edition hitps:/www.eduskunta {Fe toaeduskunnasta/jukaisuvDocuments/tuv|_11+2014 patis a global need, and therefore an actual or potential global market. We believe that the needs underpinning each GVN are significant and will be even more significant in the future. Thus, collectively, our GVNs are likely to represent an important part of the global economy in 20 years’ time. 2 Overview methodology ‘The approach of this study is based on a methodology that was first developed in the context of the Finnish Radical Technology Inquirer (RTI). The first step was to create descriptions of future GVNs from a European perspective. This was carried out through a set of interviews with 35 experts from 22 different European countries, aiming to adapt and enhance the GVN descriptions of the Finnish study to reflect an EU level appreciation of importance. The results were discussed at a workshop with experts and EU policy-makers, and were refined accordingly. In parallel, and independently from the effort to describe the GVNs, a large set of potential technological innovations was identified through a procedure that combined machine learning algorithms and human evaluation. After successive waves of refinement, a llist of 100 emerging topics has resulted. This iterative process, which is described in depth in section 4, is visualized in Figure 2. eRe Figure 2: Process of RIB Selection Simultaneously, a set of potentially disruptive social innovations were identified through a meta-analysis of existing mapping exercises.” ? Linturi, Risto, Osmo Kuusi and Toni Ahiqvist (2014) 100 Opportunities for Finland and the ‘World, the Radical Technology Inquirer (RTT), the English edition btpsy/www_eduskunta {VF tietoaeduskunnastajulkaisu/Documents/tuy)_11*2014.paf > The EU Projects CAS! http://www.casi2020.eu/ and TEPSIE http://w. tepsie. ew, OBSERVE bies./imwwnhorizon-observatory.eu and the German BMBF Foresight Il hitps:/imew.zukunft-The resulting technological and social innovations were then submitted to further assessments based on: 1) A consultation of experts from Europe and beyond, in addition to a panel of PhD students specialised in the RIB areas; (for details see Part I, section 1.3) 2) An analysis of the patents and publications; (for details see Part Il, section 1.5) By combining the outcomes of these elements, we formed three key indicators for each technological RIB and three for each societal practice RIB. Table 1 provides an overview of the sources behind each indicator. The details on the calculation of the indicators are provided in the Final Report, Part II, section 1.7 (maturity), section 1.8 (EU position), section 1.9 (likelihood). In addition, we analysed the relationship between RIBs and GVNs. In the Final Report, Part Il, chapter 2, we describe the outcomes of this analysis, which allows some conclusions to be drawn in regards to the impact of the RIBs on future value creation and societal needs. Table 1: Forming of the RIBRI indicators aitnatiania aia | sow | Likelihood of reaching market Online expert consultation '* PhD panel assessment or significant use by 2038 * Scientometric Analysis (Level and Current Maturity Dynamics of Publications & Patents) ‘© Expert Consultation, Eiropesn (postion in fiat ‘© Scientometric Analysis (position of EU in Publications & Patents) Likelihood of significant ‘expansion in Europe (RSB- © Expert consultation only) Likelihood of significant ‘expansion globally (RSB. * Expert consultation only) 3 How to read the report Chapter 4 presents the important technological RIBs that form the main body of this study. For presentation purposes, they have been clustered in thematic groups. Most RIBs are combinations of different elements, which form various directions of potential development. These are also described under each RIB. Chapter 5 presents the 13 emerging important societal practices or "Radical Social Innovation Breakthroughs (RSBs)” that we identified. These practices have implications for the emergence of future value networks, as well as forthe direction of technological innovations, and can be supported or counteracted by policy. Chapter 6 presents the analysis of the societal and technological RIBs. In particular, we assess their likelihood of significant future use, their current maturity and Europe’s position in providing policy conclusions and recommendations, which are presented in the concluding chapter. Chapter 7 presents the 23 descriptions of Global Value Networks (GVNs) used to assess the future importance of the innovations identified. We consider these to be possible future scenarios for important future global value creating structures. While we made considerable effort to create independent scenarios, there are at least two elements strongly present in many of the scenarios. One is the need for sustainability and progress towards the Sustainable Development Goals. The other is the effects of information technology in general, and the combined effects of ubiquitous networks, data and artificial intelligence. We think that these are fundamental characteristics of how value will be created in the future and that they will combine in paradigmatic configurations such as those described by our GVNs. Finally, in chapter 8, we attempt a bird's eye view on all these elements: of possible future techno-economic paradigms and venture into some overarching observations. We conclude with chapter 9. The final report consists of two parts. Part I gives an overview of the findings and describes the Radical Innovation Breakthroughs in detail. In Part Il, we give full details on the individual methodological elements such as the components of the online expert consultation and the approach to scientometric analysis. Readers wishing to go deeper into the findings can refer to the raw results from the consultation, the scientometric analysis and the impact assessment in Part I. 4 Radical Innovation Breakthroughs (RIBs) in technology In the following sections, we report the identified RIBs in eight thematic groups. The groups emerged from an automatic clustering based on a semantic analysis of the RIB texts. For each RIB, we first show a graph with the values the RIB achieved in the assessments we carried out. Then, the text presents first an overall description of the RIB followed by the more specific directions where the sources claimed recent progress. The meaning of the indicators is as follows: + Likelihood to reach the market or significant use by 2038: This indicator characterises the expectation of experts regarding the future development of this RIB over the next twenty years. For the social innovations, this indicator was replaced by Likelihood of significant ‘expansion of this practice in Europe/the world. + Maturity: This indicator describes the current status of the emerging technology. Low maturity (1) indicates first proof of concept or even first speculations, and very high maturity (5) suggests that a technology is already applied in first products. The assessment rests on the review of related patents and publications. Low maturity hasalso been assigned in cases where there is already a ihigh level of patents but, a recent steep rise in publications and patents, suggests that important new directions may emerge. We did not apply this indicator to the social innovations. + European Position: This indicator captures the strength of Europe's current capability in research and innovation regards to a technological RIB. Its assessment is based partly on expert assessment from the consultation and partly on the EU's share in patents and publications. We did not apply this indicator to the social innovations Table 1: Overview Technology RIBs (alphabetical order) Printing & Materials Printing & Materials Printing & Materials Printing & Materials Printing & Materials Energy Energy Biomedicine ‘Artificial Inteligence and Robots Energy Human-Machine Interaction & Biomimetics ‘2D Materials| ‘30 Printing of Food ‘3D Printing of Glass ‘30 Printing of Large Objects ‘4 Printing ‘Airborne Wind Turbine ‘Aluminium-based Eneroy Antibiotic Suscentibilty Testing ‘Adtificial Inteligence ‘Adtficial Photosynthesis ‘Adtficil Synapse/ Brain ‘Asteroid Mining ‘Breaking Resource Boundaries ‘Avomented Reality Artificial Intelligence and Robots ‘Automated Indoor Farming ‘Artificial Inteligence and Robots ‘Biodegradable Sensors Biohybrids Bioelectronics Biohybrids ‘Bioinformatics Biohybrids ‘Bioluminescence Energy ‘Bionics (medicine) ‘Human-Machine Interaction & Biomimencs ‘Bioplastic ‘Bioorinting (of human parts) ‘Blockchain| ‘Brain Functional Mapping ‘Brain Machine Interface: (RMI) ‘Carbon Capture and Sequestration ‘Carbon Nanatsbes ‘Chathots ‘Computational Creativity ‘Computing Memory ‘Control of Gene: Expression ‘Desalination Drivertess: ‘Drug Delivery [Breaking Resource Boundaries Biomedicine ‘Artificial Inteligence and Robots Human-Machine Interaction & Biomimetics Human-Machine Interaction & Biomimetics ‘Breaking Resource Boundaries Electronics & Computing ‘Artificial Inteligence and Robots ‘Artificial Inteligence and Robots Electronics & Computing Biomedicine ‘Breaking Resource Boundaries ‘Artificial Inteligence and Robots Biomedicineory 4 S Fa FI 3 re near ‘Human-Machine Interaction & Biomimevcs Energy ‘Artifical Inteligence and Robots Electronics & Computing ArtifiialInteigence and Robots Blomedicine Blomedicine Biomedicine Breaking Resource Boundaries Electronics & Computing Electronics & Computing Enersy ‘Artifical Inteligence and Robots Artifical Inteligence and Robots Printing & Materials Energy ‘Breaking Resource Boundaries Artificial Inteligence and Robots Biohybrids Enersy Printing & Materials Enersy Biomedicine Blohybrids Enersy Electronics & Computing Electronics & Computing Human-Machine Interaction & Biomimetics Artificial Inteligence and Robots Electronics & Computing Biohybrids Breaking Resource Boundaries Artificial Inteligence and Robots Electronics & Computing Electronics & Computing Biomedicine Biomedicine Printing & Materials Human-Machine Interaction & Biomimetics Energy ‘Artificial Inteligence and Robots ‘Artificial Inteigence and Robots Electronics & Computing Breaking Resource Boundaries ‘Artificial Inteligence and RobotsCnr ory ‘Targeting Cell Death Pathways ‘Biomedicine ‘Technologies for Disaster Preparedness Breaking Resource Boundaries Thermaslectric Paint Energy Touchless Gesture Recognition ‘Artificial Inteligence and Robots ‘Undenwater Living ‘Breaking Resource Boundaries ‘Warfare Drones ‘Artificial Inteligence and Robots ‘Wastewater Nutrient Recovery Breaking Resource Boundaries, ‘Water Solitting: Energy Most of the 100 RIBS and RSBs are collections or ‘baskets’ of related developments. Since the RIBs were assembled primarily based on news reports, which usually describe one single piece of technology, the main challenge was to group the right technologies under the appropriate headings (the RIB"title’). Generally, two broad approaches were used for this grouping: RIBs consist of developments related to each other in terms of either the science and technology they are based on (e.g. hydrogels) or in terms of the goals they serve (e.g. drug delivery). One consequence is that a piece of technology occasionally features in two or more RIBS. For example, hydrogels have a dedicated RIB due to the number of diverse solutions using this type of material, but also appear in the context of efforts to build soft robots or to improve the delivery of therapeutic substances (For more details on designing the RIBs, see Part II, section 1.2.). In drafting the RIBs, experts worked, mostly with tech news reports. Since we make no claims as to the originality of the description of these technologies, we have cited our sources copiously. Where we felt it was necessary we also placed text between inverted commas.41 Group 1. Artificial Intelligence and Robots 4.1.1 Augmented Reality Likelihood of significant use by European Position 2038 Figure 2: RIB Score of Augmented Reality Augmented Reality (AR) means overlaying computer generated imagery (or even sound) on our perception of the real world. From a technological perspective, augmented reality is a big challenge as it implies using a complex suite of sensors that can track the user position and points of attention and a good understand of the 3D environment. Currently, the mobile phone acts like a window through which the user can see the world with virtual objects overlaid on top of it, but augmented reality goggles allow the user to see the world through a transparent display that does not redraw the world. Such goggles are expensive and cumbersome but iterative development is improving them every few years. Recent progress directions Synchronization with the physical world The cornerstone of performing AR is the way the virtual projection integrates with the real world. It may sound simple, but it is not easy to completely overlap the human sensory system. Eyes must be tracked to compensate for movement.‘ The world must be scanned in 3D in order to synchronize* the virtual world with it. Some companies are using special cameras for this", but others are focusing on using commodity smartphone cameras to accomplish the same task while sacrificing performance.’ The device must also track people moving through the world.’ While people tracking augmented reality with facial recognition is controversial from a social perspective, the technology continues to be improved in this direction and police are already deploying it.’ * tps /www.theengineer.co.uk/sensor-tracks-eye-movements-in-realtime-to-enhance-virtual- reality * https:/www.technologyreview.com/s/60209 1/baidu-s-bringing-inteligent-ar-to-the-masses * https //campustechnology. convarticles/2017/07/24/stanford-team-develops-4d-camera-for-use- in-robots-vr-autonomous-cars. aspx ” http:/hwww.3ders.org/articles/20161215-korean-government-announces-affordable-34-video- capture-technology-for-smartphones html * http:/mww.dailymailco.uk/sciencetech/article-4001734/The-Shazam-taces-World-s-facial- recognition-smariphones-identiy-faces-person-print-TV.him! * hntp:/labenews.go.comiinternational/police-china-wearing-facial-recognition- glassesistory?id=52931801Live instructions The most promising professional application for augmented reality are interactive manuals giving live instructions to people working directly with a machine. In order to enable such applications, the AR device must be able to pick up objects from the world and turn them into virtual objects in real time.”® Instructions can then be projected in relation to these virtual simulacrums. Car companies are already exploring the concept, but the most novel research is happening in the field dealing with the most complex machines of all: human medicine. Doctors are using augmented reality to better guide them during surgery", and it is hoped that this will drastically reduce the time spent in the operating room. Therapy Improving the health of long-term sufferers is a welcome application for any new technology. AR has already proven that it can help people with sensory issues such as phantom limb patients.” This is successfully replacing bulky mirror systems that did not even work for all cases. Another application is to complement the missing sensory processing of autistic children’? by directly showing them hints of what they should do depending on the social context. Recuperation protocols can be improved by showing patients virtual live models of their own movements" allowing them to self-correct. Displays Forming an image very close to the eye is not an easy task. While some researchers try to miniaturize conventional very high-resolution displays, others are focusing on guiding and projecting the light rays” in such a way that a clear image is built inside the eye. Since human eyes have many imperfections, such projection technologies can also be used to improve vision.” Augmented reality with sound hile most AR solutions generate visual output, a prototype pair of glasses incorporating focused speakers enables different apps to trigger specific audio. cues in relation to the objects in the estimated direction of the sight. ® http:/www.3ders.org/articles/20170728-purdue-scientists-developing-surfnet-ai-system-that- tums 26-mages into bé-models Rim! https:/swww.theengineer.co.uk/augmented-reality-guides-surgeons-during-spine-cranial-and- trauma-surgery * http:/ineurosciencenews.com/augmented-reality-phantom-limb-pain-5660/ * http://www. dailymail.co.uk/sciencetech/article-~4900402/Google-Giass-app-tell-autistic- childtensay hind * https:/iwww.theengineer.co.uk/ultrasound-augmented-tongue-speech-therapy http://www. dailymail.co.uk/sciencetech/article-4543270/Get-ready-MegaHD-Experts-reveal- technique screens him! http:/ifeeds.nanowerk.com/~/293934432/0/nanowerk/agwb~Engineers-invent-method-to- contfolight propagation in-waveguides.php ™ htp:/hww japantoday comicalegoryechnologyview/etina-scanning-dsplay-lasses- developed-or-weak sighted people https:/hwww.theverge.com/2018/3/12/17 106688/bose-ar-audio-augmented-reality-glasses- demo-sxow 2018Long-term perspectives The ultimate goal of the currently competing technological solutions is to have an unobtrusive device that seamlessly overlays the virtual world on top of the user’s natural senses. This device may be a pair of glasses, contact lenses or even a cybernetic implant to beam the virtual world directly into our nerves. While such devices are still far, companies are cautiously investing in this direction. For productivity, instead of being tied down to a PC monitor, a whole wall could serve as a desktop workspace. Teleconferencing would be as easy as looking across the table to the virtual representation of your colleague. For people working on site, plans for modifications would be overiaid directly in the room. No need to fumble with a manual when augmented reality can show you how to fix an engine by highlighting the components that need intervention. The advertising industry is taking particular interest in augmented reality seeing it as both an opportunity and a threat. There is a great benefit to showing people personalized ads in the real world, but those same people might want to install an augmented reality ad-blocker and stop seeing any billboards at all. Gaming is another great fit for the technology. No longer would computer games be tied to the living room. We have already seen a glimpse of this with the Pokemon Go mobile app. That game only implemented rudimentary augmented reality but the features garnered a large number of players. With seamless AR, every human would see his or her personalized version of the world. Imagine all what you see in your field of vision is translated into your language, or personalized signs that point you to your airplane seat, or the colour of the sky changing to fit your mood, or virtual pets following you everywhere. However, with everyone seeing things their own way there is no telling how society would change. 4.1.2. Automated Indoor Farming European Postion Figure 3: RIB Score of Automated indoor farming ‘There are a number of reasons to move farming indoors. In areas with radioactivity - such as those experiencing the aftermath of nuclear disasters ~, there are fears that traditionally grown produce could contain radioactive fallout. In other areas, a desert environment, lacking in water resources or other geographical factors, may pose challenges for growing vegetables. Insuch cases, factory farming, which is mostly indoors, may be viable and scalable. Recent progress direction Fully automated farming process Japanese companies are advancing fast in robotic indoor farming that is stable in any climate condition. More than 200 ‘plant factories’ in Japan are capable of harvesting 20,000 heads of lettuce daily.’ Their lettuce is grown hydroponically, in a nutrient-rich gelatinous substance in a sterile environment in vertical stacks under LED lights. One example is Spread, a company running “a giant factory farm that grows lettuce” in Kameoka, Kyoto Prefecture and another one at the plant factory at Kansai Science City.” The so-called vertical farms allow for a highly efficient use of water. At the same time, automation of outdoor farming is also progressing quickly. Guided by artificial intelligence systems, machines carry out classical farming tasks such as raising seedlings, replanting and harvesting but also animal husbandry. Techno farming in extreme conditions The concept of indoor factory farm is now being exported to the Middle East, where the desert environment, lack of water resources and other geographical factors pose challenges for growing vegetables. Long-term perspectives Some people think that, in the long run, agriculture could become fully automatized, first in areas with a lack of human workers and extreme conditions and then around the globe. This could have disruptive impacts areas in like food culture, sustainability, social fabric and, last but not least, employment. ™ https:/www japantimes.co_jplife/2017/03/10/food/indoor farms-next-step-evolution- agricutturet# Wp8CCEOmS2s * https:/iwww japantimes.co.jplife/2017/03/1O/food/indoor farms-next-step-evolution- agricutture# Wp8CCEOmSs2s41.3 Blockchain Lxethood of signfcant use by 2028 European Postion Figure 4: RIB Score of Blockchain The blockchain is a technology that allows people who do not know each other to organize a network to keep trusted records. Once recorded on a blockchain a piece of information cannot be changed. The most common application is that of a “distributed ledger” that keeps track of transactions. This solves, among others, the problem of double spending digital money (or other formation). Using the blockchain, digital data can have an owner recognized by everyone in the network. The blockchain is also the technology at the core of cryptocurrencies like bitcoin. The way this network functions is a matter of debate and different implementations have sparked new ideas and development directions outside the original scope of the concept. At its core, a blockchain is a data structure consisting of a linked list that is cryptographically secured so that its elements cannot be altered after they have been set. Unfortunately, it also comes with a cost, as the energy consumption for Bitcoin mining is estimated to be comparable to that of Ireland and increasing.” Recent progress directions Trust and notarization As data registered on the blockchain is immutable, data (“hashes”) in any document can be stored there and verified at any point in the future. This is bringing considerable disruption to the field of notarization by removing the human intervention from the loop and providing anonymity.” Companies are terested in storing the data about their product in immutable databases.” This way they will be able to reliably trace any information about their * https.//arstechnica.com/tech-policy/20 1 8/05/new-study-quantifies-bitcoins-ludicrous-energy- consumption’ https:/iguce oath com/collectConsent?brandT ype=nonEu8. done=https%3A%2F %2F techcrunch ‘62F 2015%2F09°%2F 22°%2Fstampery-leverages-the-blockchain-to-certify-al-your- documents %2F®%.3F guccounter%43D 1&sessionid=3_cc-session_69504575-d56c-4d49-a015- 531Ge3d!14758lang=Binline=false * https: www. wired.com/201 7/05icurious-plan-save-environment-blockchain!products when queried.* This has the potential to improve operations considerably and reduce the legal burden considerably. Smart contracts Code that transacts the digital assets is being made to be self-executing by tering it on the specific blockchain itself.* This means that any transaction that required escrow can now use a smart contract instead. Furthermore, any money processing operation (exchanging currency, routing or even gambling) can be implemented as a smart contract. In practice, every member of the blockchain network gets a readable copy of the code. The validating nodes of the blockchain network then execute the stored code when the prerequisite conditions are met. Transactions executed by smart contracts need to be done in cryptocurrency (Bitcoin, Ethereum and so on). It is hoped that this use case of cryptocurrencies will further legitimize their existence in the financial sector. Private corporate blockchain networks Not all networks have to be public and as such business and IT companies are investing considerable resources in developing private blockchain networks””* that can be deployed as easily as cloud infrastructure for any need a company might have. These private blockchains could be used for storing data, certification or even some tightly controlled form of cryptocurrency. Proof of Stake Protocol Currently, the main method for “proof of work”, in which the miners’ competition involves a process of generating signatures repeatedly until one guesses correctly a random number, the consumption of energy being mense.** An alternative method called “proof of stake”, which select validators based in part on the size of their respective monetary deposits (their stake) is currently under review to be adopted by Ethereum, one of the leading cryptocurrency network.** Long-term perspectives A blockchain-enabled world without central controll happen through decentralized networks, providing a neutral and fair result to > http:fortune.com/2017/05/30/walmart-blockchain-drones-patenll * https:Jwmw.coindesk.com/ethereum-blockchain-homestead * hittps:/cointelegraph.com/news/trust-but-verify fist-ethereum-decompiler-launched-with-jp- morgan-project hittp:ortune.com/20 16/10/04/jp-morgan-chase-blockchain-ethereum-quorum! * hittps:/www.prnewswire.com/news-releases/chronicled-launches-quorum-blockchain- integration-at-enterprise-ethereum-aliance-kickoff-300414641 html hittps:/www.coindesk.convibm-goes-big-on-blockchain-unveling-ambitious-new-service- offerings-and-strategy! > https:/Recherunch.com/2017/08/10/microsoft-wants-to-make-blockchain-networks-enterprise- ready-with-ts-new-coco-framework/?guccounter=1 > hitips:/hmww.technologyreview.com/s/609480/bitcoin-uses-massive-amounts-of-energybut- theres-a-plan-to-fixi! hitpsJ/eryptoslate.com/ethereums-proof-of-stake-protocol-in-review!le transactions. Presently, companies are seeing blockchain technology as an opportunity to increase traceability in their own business. Perfect immutable records can be kept without any hassle or risk of contamination. They can be verified at any time by anyone on the network. This can be used to boost the transparency of an endeavour. As such, the blockchain is accepted by public groups and private companies as a future infrastructure for honest business. Most if not all leading companies in technology are interested in deploying blockchain implementations relevant to their fields. 4.1.4 Chatbots European Figure 5: RIB Score of Chatbots Chatbots, short for “chatting robots”, are computer programs which conduct real-time conversations with humans, via written text or live au Designing a chatting machine that would be as adept at conversations as an ordinary person has fuelled computer scientists’ imagination and ambition ever since 1950, when British mathematician Alain Turing framed that challenge as a famous benchmark for artificial intelligence. Until recently, software research around chatting programs only progressed at a slow pace and did not yield notable results. During the last years, however, significant software and hardware advances in natural language processing, machine learning, and cloud computing have given rise to a host of widely available chatbots, employed for personalized (albeit still limited) conversations. Some chatbots are standalone applications. Others are integrated in messenger applications, social platforms and larger IT ecosystems, on a range of digital devices. While many chatbots are aimed at assisting with specific tasks (such as placing shopping orders or finding out the weather forecast), the range of chatbot capabilities and uses is quickly expanding, spanning areas such as banking, telecom and entertainment. Recent progress directions Unscripted, AI-powered chatbots Traditionally, chatbots have followed a set of predefined rules and scripts, looking for certain words and providing predefined answers to predefined questions. This paradigm has often led to an underwhelming user experience. Newer chatbots are powered by Artificial Intelligence technologies (such asNatural Language Processing and Machine Learning), allowing them to be more flexible in terms of user input and blurring the line between chatbots and virtual assistants like Siri, Cortana or Google Assistant. Machine learning techniques (and, in particular, Deep Learning models) enable the automatic training of chatbots, mitigating the issue of question variants. Businesses can program Al-powered chatbots to pull information from a trove of available data and self-improve over time: the more they are used, the more they are learning and the better they are ge 9. The end result is that an unscripted chatbot has fewer limitations: it is far more likely to understand a question and to provide a helpful answer when a response does not fit a narrow script. Even when a chatbot is unsure about what the user exactly asked, it can attempt to offer multiple options rather than defaulting to the standard (and dreaded) response: “I don’t know”. Technology reuse and integration with major platforms Most major high-tech players have invested significant resources in the development of chatbots and virtual assistants. Having developed competing technologies, the big consumer-facing IT companies have integrated chatbots and virtual assistants in their messaging platforms and ecosystems. Besides deploying chatbot services directly to end users, the companies have also opened up their conversational technologies to outside developers. In April 2016, Facebook announced support for chatbots in its very popular Messenger system™, leading to a flurry of Messenger chatbots launched by third parties (jumping from 30,000 to 100,000 within a single year), alongside notable hype. One can use dedicated chatbots to book an appointment or to order a Uber ride, directly in the Messenger platform.** Google’s 2016 Allo, a smart instant messaging app, allows people to chat directly with Google Assistant and receive, for example, restaurant or movie suggestions. Besides integrating Cortana, the Al assistant, into its core products, Microsoft launched Bot Framework, a set of programming tools for chatbot developers, in early 2016. Over 130,000 developers have since registered for the service. rosoft’s two chatbots for Chinese and Japanese users, Xiaoice and Rinna, launched in 2014 and 2015, have been used by millions as information providers - but also for interaction at a personal and emotional level, perhaps in part due to techno-cultural prociivities. Apple and Amazon are also pushing further the development of their own smart assistants, SIRI and Alexa, which embed chat functions. In April 2017, Amazon made Amazon Lex - a set of conversation tools also used by Alexa - generally available to other companies. In Europe, Telia, Scandinavia‘s largest operator of mobile, fixed and broadband services, announced in the summer of 2018 the launch of A Conservational Engagement (ACE). This customer relations platform adds the * http:/iwww.information-age.comlchatbots-more-chit-chat-123469420 ™ There were 1.2 billion registered users in mid-2016 (https:/getvoip.com/blog/2017/04/21/the- current-state-of-chatbots-in-2017) * https:/igetvoip.com/blog/2017/04/2 the-current-state-of-chatbots-in-2017latest AI technologies to a pre-existing chatbot ecosystem already in use elsewhere on the continent (such as by electricity giant E.ON).”° Enterprise and Customer Service Applications Chatbots are increasingly used for customer care across a spectrum of fields, \cluding commerce, banking and telecom. In customer service, the focus is on solving a customer query, rather than on personalization and good conversational skills. Generally, chatbots are complementing and assisting customer service representatives (CSR) in order to reduce workload, rather than replace the latter. They are mostly used to handle routine questions, retrieve information from technical databases, and redirect calls or inquiries. The financial sector is increasingly testing and deploying chatbots. While legacy banks seek to supplement their core products, a number of fintech startups have built chatbots as core products. Supported transactions include balance checking, bill paying and money transfers, such as the daily transfer of small amounts into savings (microsaving).” While traditional banks are somewhat slower to adapt, they do not sit still either. In Scandinavia, banking giants SEB, Swedbank and Nordea have introduced Aida, Nina, and Nova, respectively, in 2018, after several years of internal tests. These Al-powered assistants have emotion-sensing and other state-of-the-art capabilities. Across the oceans, Australian giant NAB has developed a virtual banking assistant powered by AI which is able to answer more than 200 common questions relating to business banking accounts, with 13,000 variants, extracted from real-life customer enquiries. When the bot cannot answer a question, a human banker takes over.” In early 2017, a survey undertaken in Australia suggested consumers were warming to bots, although 56 % still preferred to interact with a human.“ Long-term perspectives While a general-purpose chatbot on par with humans is not expected in the foreseeable future, chatbots will likely evolve and enter the mainstream, as they are getting progressively better at understanding and responding to user questions and commands. As chatbots get smarter and smart assistants get chattier, the line between chatbots and smart assistants will further blur. ‘The future of chatbots (as well as virtual assistants) is linked to the future of Artificial Intelligence. Assuming continued advances in AI algorithms and chips (and, particularly, in NLP), AI powered chatbots will get better and better at understanding the intent behind a human’s question, as well as at providing meaningful answers in a variety of conversational scenarios. Future chatbots could lead to rich conversational user interfaces, allowing users to * https:/iwww teliacompany.comleninews/news-articles/201 8/the-creators-of-a-grand-new- world! * hitp:www.businessinsider.com/chatbots-banking-ai-robots-finance-2017-10 * https:/www.computerweekly.com/news/252434043/Swedish-banks-embrace-rapidly- evolving-chatbots * htips:/iwww.cmo.com.aularticle/626835/nab-taps-power-ai-chatbots-business-customer- service * hittps:/www.crno.com au/article/6 19915/report-aussies-happy-talking-chatbotsnaturally interact with a trove of devices (including computers, smartphones, robots, smart homes) rather than using all kinds of graphical interfaces and applications. 4.1.5 Computational Creativity European Position Figure 6: RIB Score of Computational Creativity Computers are now capable of producing output that is (almost) distinguishable from human works.“ They use software, which assesses formation, identifies a gap and then uses the best elements at its disposal to create something brand new. In other words, computers are not randomly mixing existing data; instead, they use scientific methodologies and, by merging existing features, produce creative results. Recent progress direction Computers are capable of creating original art, ideas and solutions that look as much human-generated as the work appearing at major art fairs or coming out of well-established think tanks. ‘The semi-autonomous AI systems producing these works are supported by human designers. They are programmed by humans, but identify new avenues, new solutions, and new ideas by having no preconceived limitations and by using massive amounts of processing power. Alongside Machine Learning and Natural Language Processing, as well as computational perception and contextual awareness®, Computational Creativity is an integral part of finding solutions. "The AI can explore a space, identify things that are potentially interesting, and then present them to a designer to fallow up on.“ Some machines can also teach themselves - a special AI (AutoML) can now pick out specific objects in images better than any other computer vision hittps:/news.artnet.com/art-world/rutgers-artficil-inteligence-art-1019066 © ttp:/telecoms.com/486958/nvidia-software-raises-question-as-to-whether-creativity-actually- exists! * https-Jwww.technologyre view.com/s/609482/ai-is-dreaming-up-new-kinds-of-video-games/system.“ This is an AI form effectively created by another Al, on rules and large swaths of data. g only In some domains, “ignorance of tradition and precedent is a clear weakness for an Al system”, but in finding new designs it represents “a strength that could unlock new creativity. Moreover, it could help managers and designers lower the cost of producing new strategies and products”. Long-term perspectives Creativity through AI play an ever-increasing role in the future, augmenting humans in their quests and problem solving, besides the automation of menial tasks. The next frontier is using increasingly sophisticated machine-learning techniques to design entirely new kinds of objects or strategies that have, to date, evaded the human imagination. 41.6 Driverless Eaopean Posiion Figure 7: RIB Score of Driveriess For a long time, the driver was thought to be indispensable for any type of vehicle. While an autopilot that can handle long monotonous sections without obstacles was considered standard for water and air vehicles, no solution was available for land vehicles. Autopilots also failed completely when encountering even small disruptive elements. Thus, the driver remained central to the operation of any vehicle. Even fiction did not stray from this idea too much, as most visions of the future lacked any truly self-driving vehicles. This all changed when the DARPA grand challenge proved that driverless vehicles were not only possible, but within our grasp. Surprisingly, self-driving technology was not descended from autopilot systems. Instead, it was an extension of autonomous robotics research. For decades, roboticists had developed many algorithms for autonomous robots without much thought of practical applications outside the lab. Blue-sky research into small (person-sized) robots, which could roam around on their own, turned out to be relatively easy to scale up. Turning a traditional car into a self-driving car “ http://www.ifiscience. com/technology/googles-ai-makes-its-own-ai-children-and-theyre- awesome! htps:/www.technologyreview.com/s/600482/ai-is-dreaming-up-new-kinds-of-video-games! * hitps Jhwww.technologyre view.com/s/609482/ai-is-dreaming-up-new-kinds-of-video-gamesboiled down to equipping it with enough sensors and computers to essentially turn it into a large autonomous robot. ‘The winners of the DARPA grand challenge were not perfect. In fact, their Performance was amusing sometimes. However, they proved to the world that it could be done. Suddenly everyone could envision a world where the driver is not a mandatory part of any moving vehicle. The world braced for the inevitable appearance of self-driving cars and the extinction of driving as a requirement (or even a job) but the revolution has been reluctant to make itself felt. It turns out that all the sensors and computers needed to make a car self- driving are very expensive. Not only that, but they are very fragile. Nobody wants to equip a car with electronics worth many times more than the car itself and replace the failing ones in a few weeks. Furthermore, real world behaviour of human drivers appears to be hard to anticipate as it varies wildly from day to day and region to region. This set of problems refocused research on creating self-driving vehicles with fewer, cheaper sensors and making the vehicles themselves capable of learning. These efforts are intertwined, as the most efficient way of making a self-adjusting system with few sensors is by making it capable of learning. Unfortunately, this brings us into uncharted territory. "Robodrivers” resulting from learning algorithms are not fully understood even by their designers. Some experts are quick to raise the alarm and deem them unpredictable. Questions have arisen about how self- driving vehicles would handle ethical dilemmas. Although much safer than human drivers, such vehicles might accidentally injure or kill someone. Engineers believe this is a non-issue because systems prefer inaction to making moral choices, but critics are not convinced. Despite the criticism, industry is moving ahead. Every major company volved in vehicles (land, sea, air or any other kind) is investing significant resources to bring this technology comfortably to market. While the public deployment of the technology will not be as sudden as originally thought, it slowly start to change things. The long-term effects are still undetermined. As the transportation industry changes, it usually induces significant change in alll other fields connected to it. Recent progress directions New-generation sensors Since one of the main obstacles to wide availability of driveriess technologies is the relative cost and complexity of sensors, a lot of effort goes into finding new ways to perceive the world. Modulated electromagnetic waves are already omnipresent in our world so maybe they can also be used to guide” driverless vehicles. The general idea would be to cover the blind spots and imperfections of GPS and other traditional navigation signals so that fast moving driverless signals do not have to always rely on them. Instead of building a perfect representation of the environment in the computer memory, several companies are trying to train the computer to think using more limited data. Car companies will try to outfit regular cars hhttp:/www.azosensors.com/news.aspx?newsID=11583%20with enough sensors to build a rough understanding of the driving environment* and then train a computer® to link the environment to what the human driver is doing. Even the “traditional” LIDAR (laser radar) is being augmented™ in order to see through foliage and bad weather. Some are looking to replace LIDAR completely with special cameras* that capture “light fields”. These light feed cameras operate passively and absorb so much information about the light rays entering them that a computer can reconstruct the 3D environment the camera films. Another novel approach involves turning the sensors inward to allow driverless vehicles to sense themselves and predict breakdowns.” Man-machine synergy From an interface design point of view, driverless vehicles are unexpectedly complicated. If the vehicle is expected to carry humans, it must communicate with them in an intuitive and clear way. This means outfitting the vehicle with internal sensors dedicated to passenger analysis. The problem is compounded if there is a fall back system allowing a human driver to take control of the vehicle. Direct brain interfaces have been explored as a solution™ allowing the human to direct the vehicle while the vehicle can monitor the mental state of occupant. Even i not expected to carry humans, a driverless vehicle must be able to interact with humans it may encounter in the environment. To this end, algorithms are being developed that can analyse humans instantly in order to understand their movement and intent. Legislation Many developers see legislation as an obstacle to the development of driverless vehicles, but many jurisdictions have proven to be supportive rather than disruptive. When it comes to land vehicles, it has been decided that the definition of driver is outdated™, as it does not describe who may be in control of the car. California, with its many tech companies, has gone a step further by allowing driverless cars to operate without any human inside”, while making the owner of the vehicle responsible for the actions of the computer in control of the vehicle. “ http://www .techradar.com/us/news/car-tech/oyota-plans-to-use-self-driving-car-cameras-gps- to-make-more-accurate-road-maps-1311730 © https:/www seeker. com/a-way-for-driveriess-cars-to-see-the-road-and-react- 1770642552.htmi ® https:/www.eurekalert.org/pub_releases/2017-06/tos-stf0627 17.php *'https:/icampustechnology.com/articles/2017/07/24/stanford-team-develops-4d-camera-for- use-in-robots-vr-autonomous-cars.aspx © https:/www_springwise.com/startup-using-ultrasonic-sensors-revolutionise-automobile- industry! © hitp:/www.azosensors.com/news.aspx?newsID=12074 * http //venturebeat.com/201 5/12/07/chinese-researchers-unveil-brain-powered-car! * https:/www.eurekalert.org/pub_releases/2017-07/cmu-act070617.php * http/irobohub.orginhtsa-redefines-driver-of-self-driving-car! * hitp:/irobohub.org/california-proposes-driverless-car-rules!The high seas continue to enjoy a special status even in this new age. There is currently no legislation governing autonomous ships™ and companies are eager to take advantage of this Connectivity When a vehicle is continuously moving through a landscape at high-speed, connectivity becomes a true problem. Traditional data connections become hard to maintain and the driverless vehicle cannot broadcast or receive the data it needs. State of the art telecom infrastructure is being developed with this in mind. The complete opposite approach is to have many autonomous vehicles communicate with each other“ to create a mesh network. In this way, only a few vehicles at a time would be able to relay all the data to the outside world. Of course, the last resort is to develop a vehicle that can operate completely autonomously“, but even such a vehicle would benefit from a means to connect reliably to the internet if needed. As for the occupants, turning the vehicle itself into a communication hub would mean the devices inside will connect to it using standard protocols, while the special antennas of the vehicle handle outside connections. Long-term perspectives The concentrated effort to create fully autonomous driverless vehicles continues. However, despite the transnational resources devoted to developing the technology, the vision is not as clear as many originally thought. The promise of having wide scale deployment of completely autonomous driverless vehicles by 2020 is unlikely to be realised. Instead, companies are focusing on creating almost-autonomous vehicles that can safely let a human take control during unforeseen situations. This might be an attractive option, but several groups have pointed out that people will slowly be deskilled by self-driving systems. As a consequence, humans might not be able to react better than a computer. In the commercial sector it is precisely this conundrum that is keeping jobs moderately secure. Professional drivers are still expected to perform better than computers. One can envision a future in which self-driving vehicles handle most of the tasks while human drivers are available to step in whenever needed. In the long term, once completely self-driving vehicles become the norm, ty will start to experience paradigm shifts. Workers will be able to use the time usually wasted on commuting. Private car ownership might not be attractive to many people anymore. Transportation is set to become a commodity regardless if it is by land, air or sea. Businesses owners envision many possibilities. Food trucks may roam the city serving food by app. Innovative package delivery systems would combine trucks and drones to ® https//singulartyhub.com/2017/07/30\the-worlds-first-autonomous-ship-wil-set-sait-in-2018) ® hittp:/www telecomstechnews. cominews/2017/feb/22/59-edges-closer-china-mobile- announces-large-scale-testing! © htipy/www.aerotechnews. com/blog/20115/12/02/lockheed-martin-conducts-collaborative- unmanned-systems-demo/ © ittps:!Nechcrunch.com/2017/02/21/exyn-unveils-a-to-help-drones-fly-autonomously-even- indoors-or-off-the-grid/?ncid=rss © hitpz/techcrunch.com/2015/12/23imeet-kymeta-the-company-that-could-bring-high-speed-wi- fito-cars!Sponsored by OKTA