Content

5. Analysis ofAl in the Information Economy in:

Henry Alexander Wittke

Artificial Intelligence, page 34 - 50

An Approach to Assess the Impact on the Information Economy

1. Edition 2020, ISBN print: 978-3-8288-4459-9, ISBN online: 978-3-8288-7480-0, https://doi.org/10.5771/9783828874800-34

Tectum, Baden-Baden
Bibliographic information
all, this methodological approach will contribute to this goal by discuss ing the political possibilities of actions in the final part of the research guided by the hypotheses. 5. Analysis o fA l in the Information Economy This chapter seeks to develop a framework to assess how strong AI will impact the Information Economy. The conceptual approach here for con sists of the assumption that Machlup’s description of the Information Economy is still very much valid. Surely, changes have occurred (also due to the development of information technology) in the different seg ments described by Machlup. But on a macro view, the concept remains intact and the five segments of concept shall be investigated, which are: Education, Information Services, Information Machines, The Media of Communication, Research and Development. Within each of these in dustries, one sector or sub-segment will be investigated. A precondition for the investigations is that sufficient literature is available to conduct a qualitative analysis. It will be attempted to show, which and how many of Machlup’s seg ments will experience strong impacts from AI in the future. This further requires analyzing the technology deployed within the segments or in dustrial sectors and how many fields of applications for new AI technol ogies are present. An assessment of the suitability and probability of workforce replacement by strong AI is equally based on literature. The following analysis uses three expressions with the following meaning: Short-term describes a timespan less than five years, mid-term means five to fifteen years and long-term should be used as a synonym for more than fifteen years. 5.1. Education In the wake of the digitalization and AI focused automation, there are discussions on how to revolutionize the education sector with emerging technologies, whether gamifying instructional materials or expanding access to knowledge via massive open online courses. Based on recent reports the market of education technology acceleration will be hugely dictated by the global education expenditure market, as education be comes increasingly more expensive. In fact, the market stands now at over five trillion dollars, 8x the size of the software market and 3x size 34 of the media and entertainment industry, yet education is only two per cent digitized. Education technologies, broadly defined as the use of computers or other technology to enhance teaching, are becoming a global phenomenon, and as distribution and platforms scale internation ally, the outlined market is projected to grow at 17 percent per year, up to 252 billion dollars by 2020. Even private investment in educational technology grew 32 percent annually from 2011 through 2015, rising to 4.5 billion dollars globally. A I’s share of these flows is likely to increase because especially strong artificial intelligence technologies might be well suited to achieve crucial education objectives, such as enhancing teaching efficiency and effectiveness, providing education for every body, and developing the skills that will be essential in the 21st century (cf. Newswire Europe 2016; McKinsey 2017: 65). Education as one of the primary industries of the developed in formation economy is already distinguished in the book „The production and distribution of knowledge in the united states“ by Fritz Machlup in the 20th century. The exact characteristics and main drivers of the indus try changed significantly in the last decades, but it can be assumed that the basic categorization of the information economy itself is also appli cable in modem days. By that one important subindustry is still job edu cation, which seems very important concerning the knowledge worker explained in chapter three. Usually, analysis requires further distinctions. With regard to the production and acquisition of knowledge while work ing on a job, for example in the case of an employee of a business firm, one should distinguish between training on the job provided by the em ployer and learning on the job on the part of the employee without any training activities arranged by the employer (cf. Machlup 1962: 51f.). Furthermore, higher education seems to be important to train the knowledge worker constantly. Education in jobs and higher education should be the analysis object within this chapter. After a short introduction, the focus will be on how the different sub-industries of education could develop with emerging possibilities of strong AI. Following recent reports by Pearson in collaboration with Univer sity College London Knowledge Lab or McKinsey, the impact of AI on the education sector in the future will be significant. They note that to day's model-based adaptive systems are increasingly transparent, allow ing educators to understand how a system arrived at a next-step decision 35 and rendering them more effective tools for classroom teaching (cf. Faggella 2017). The first important aspect to look at, are virtual tutors powered by Al, which could possibly disrupt the education sector in the future. Strong Al is supposed to personalize and optimize teaching and learning at all. So-called adaptive learning solutions aim to address the limitations of conventional classroom teaching. Accordingly, personalizing lesson plans to the student or worker existing knowledge, particular learning preferences, and individual progress could be methods. Instead of deliv ering one lesson to an entire group, which can leave behind struggling participants or disengage fast learners, adaptive learning claims to deliv er the right content, at the right time, in the best way to every single per son. Al could easily improve adaptive learning and personalized teach ing by identifying factors or indicators of successful learning for every single person. Not only monitoring such variables as the number of breaks or unconcentrated moments during a lesson or the amount of time needed to answer a question is a way to enforce the individual learning progress. In addition, computer vision and deep learning could call in new information such as mouse movements, eye tracking, and sentiment analysis, delivering a deeper insight on everybody's individual perfor mance, mindset, or cognitive ability. This data could be used to support students in real time. Implemented at scale, AI-enabled adaptive learning could restructure higher education and education on the job. By that, it could mean an end for traditional testing systems and measure academic abilities and achievement in a more nuanced way. Class formats would give more room for individuals to learn according to their preferences and needs, with teachers focusing less on lecturing and more on coaching, aided by prescriptive analytics to choose the most effective methods. Finally, the use of strong artificial intelligence at this point could empower learners by providing them with control over how fast or in which way they learn best, and the lifelong feedback of one's own cognitive and behavioral preferences. Strong Al could easily encompass all of the individual learning styles and provide a customized plan for each person after get ting to know the learning behaviour of a student and how he learns. Along with this plan, the programs imagable as a so-called individual AI-powered tutor could help to craft more specific and even more realis tic goals for every single learner (cf. Avery 2018; McKinsey 2017: 67; Lynch 2018; Turbot 2017; Maney 2017). Because educational software is adapted to individual needs, there is evidence to suggest that intelli 36 gent tutoring systems perform as well, if not better, than individual hu man tutors for many students (cf. Faggella 2017; TeachThought Staff 2017; Phillips 2018). On the one hand, deep learning algorithms could predict outcomes and prescribe accurate solutions. On the other hand, they could explain how the algorithm itself reached its conclusion and helped retro-engineer drivers of educational success. This will allow in dividual learners and workers to reflect on their cognitive abilities and understand their optimal learning setting by getting constructive feed back. Empowered by strong AI, individuals could build their own virtual robot teachers and learning partners to help them navigate lifelong learn ing experiences. The promise of strong AI at this point of consideration is to use digital tools as a cognitive window into each person's minds and help tailor learning to maximize individuals potential (cf. McKinsey 2017: 67f.; Pearson 2016: 23). Based on unleashing personalized learning also the learning expe rience will change entirely for everybody involved in education, even teaching in higher education requires a reconsideration of educators and pedagogies role (cf. Popenici/ Kerr 2017: 11). Every teacher and educa tor could benefit from virtual guidance. In an information age dominated by digital content, teachers must shift from a sage on a stage to the guides on the side. In other words, coaching might be more important than teaching in the digital age because virtual environments have al ready allowed teachers to make this shift. Coaches help students by set ting their own goals and reflect on their growth in the environment, while also setting up more lengthy open-ended projects to ensure learn ers apply the content to real-world situations. This blended environment allows human learners and machines to work symbiotically (cf. Wagner 2018). Arguing that one of the teachers most valuable asset is time, strong AI could help them to spend their time more wisely, efficient and goaloriented. Before the onset of AI, much of their time was spent develop ing and delivering content in a rigid, inflexible curriculum framework (cf. Wagner 2018). Nowadays AI is starting being able to do some of the basic activities in education and mundane duties that consume a teachers time. Also, routine grading is a task that eats up a teachers time. Strong AI can easily score hundreds of multiple-choice tests with precision ac curacy. Furthermore, it is becoming more and more possible for AI to grade written essays, or oral presentations and simplify the process of grading in this way. All in all, it will take AI far less time to do so than a 37 human. In return, the elimination of time-consuming grading tasks pre sents the opportunity for educators to pay more attention to creating a better classroom lesson or concentrating on the students and trainees themselves. This focus can be seen as a value-adding and more im portant task connected to the human teacher role (cf. Phillips 2018; Lynch 2018; TeachThought Staff 2017; Narula 2018). This requires spe cific skills, such as emotional intelligence, creativity, and communica tion, that are beyond machines current capabilities. That is why the hu man teachers should concentrate on staying good at these things. Instead of supervising and answering routine questions or taking attendance could be seen as another time-consuming administrative task of a teach er, which could be done by an AI system in the future as well. Natural language, computer vision, and deep learning could help replace teachers also in answering students routine questions or acting as tutorial supervi sors. AI could also give teachers additional support in forming the most effec tive classes by applying machine learning algorithms to data from learn ers education profile, surveys or social media. Anyway, in most cases, the upcoming improvements in the field of AI will shift the traditional role of teachers to that of facilitators (cf. TeachThought Staff 2017; McKinsey: 68f.; Houser 2017). The improvements towards virtual assistance and virtual teachers will increase significantly in the next years. Also in situations if it is not possible to have an experienced advisor, AI could be the experienced teacher, powered by learnings from big data AI can fill the gaps in sub ject areas in which a teacher does not have a particular expertise or helps by training teachers when there is a skill shortage in the job market (cf. Avery 2018; Houser 2017; Pearson 2016). With a much wider reach, AIassisted teaching could also have a significant impact in third world countries and remote locations by supporting two key enablers of teach ing, which can be described as coaching and assessing (cf. McKinsey: 68f.). Furthermore introducing AI to educational settings will benefit learners of all ages. For example for many subjects, as people get older, they are not willing to take that learning risk where they are confused. For example, Bill Gates emphasizes within this context: „The idea that you could talk to a virtual advisor that would under stand different misconceptions and arbitrary linguistics around it, that will certainly come in the next decade.” (Avery 2018). 38 Another major asset of strong artificial intelligence might be the ability to know more exact details and having higher expertise than every hu man educator. Because trial and error is a critical part of learning, not knowing the answer is paralyzing for learners or educators itself. This could change with strong AI as well. A further promising aspect is an increase in flexibility. Individuals can start learning from anywhere in the world at any time, by using AI systems, software, and support. These kinds of programs taking the place of certain types of classroom instruc tion, strong AI may replace teachers in some instances (cf. TeachThought Staff2017). The last relevant aspect of strong AI within this chapter might be taking a key role in better connecting education systems and labor markets, which could mean benefits for every single knowledge worker and the society at all. Also connecting talent with opportunities in the job mar ket, improve recruitment results or enabling education systems to meet the needs of future employers better can be seen as further advantages by strong AI. Furthermore, the technology itself could be used by govern ments to forecast detailed job-market demand more accurately and steer educational institutions to adapt their curricula and approaches accord ingly, making sure students have the skills required to fill those jobs (cf. McKmsey 2017: 65ff.). As analyzed so far, strong AI solutions will have the potential to structurally change education on jobs, higher education or education at all. As already analyzed, AI might raise its quality by making it more personalized or efficient as well for the learners as for the educators (cf. Pearson 2016: 21). However, the realm of teaching and learning in high er education, for example, might also present a very different set of chal lenges. At the moment AI solutions relate to tasks that can be automated but can not be yet envisaged as a solution for more complex tasks of higher learning (cf. Popenici/ Kerr 2017: 11). Anyway, further develop ments in the field of strong AI are important to look at. By giving an out look, the most relevant development in the next years might be the hu man-machine integration. “AI is doing some of the very labor-intensive data collection and analysis that is best done by technology, leaving the teacher to do the human interaction that’s much better done by humans (...) You keep the bit that the humans are particularly good at, and then you try and automate the support within that system.” (Avery 2018). 39 In the early future human educators still have to concentrate on things that are beyond machines current capabilities. These are for example value-adding tasks, emotional intelligence, creativity, interpersonal communication or building relationships. Not to be overlooked is the apparent fear of some scholars that human educators and teachers will be replaced by strong AI technologies in the coming decade. AI will most likely not replace most of the human edu cators in short-term or mid-term but will serve as an invaluable exten sion of the human expert, helping teachers to more effectively meet the diverse needs of many learners simultaneously (cf. Faggella 2017; Houser 2017). Another fact against the mentioned human replacement in education is, that currently, only a few percents of the education sector are digitalized, which in turn can be seen as a basis for strong AI. After the educational sector is more digitalized, the development of human replacement could speed up. It may be concluded that profound changes arising from strong AI have to be expected in the education sector, even only sub-segments have been investigated here. 5.2. Information Services Information Services mean this part of the information economy where agencies or departments are responsible for providing processed or pub lished information on specific topics to an organization's internal users, its customers, or the general public (cf. Business Dictionry 2018). Machlup describes it as an important industry for knowledge production, which is fundamental for the information society. He describes to under stand by knowledge-production any human or human-induced activity effectively designed to create, alter, or confirm in a human mind a mean ingful apperception, awareness, cognizance, or consciousness of whatev er it may be: „The activity of telling anybody anything, by word of mouth or in writing, is knowledge-production in this sense. A person ex clusively engaged in this activity belongs to a knowledge- producing oc cupation. A number of firms exclusively engaged in selling information or advice belong to a knowledge-producing industry.“ (Machlup 1962: 323). In his book „The production and distribution of knowledge in the united states“ he made a solid empirical analysis of the existing industry be cause it is one of the main parts of the information economy. The most essential subdivisions after Machlup are professional knowledge services 40 including legal services, engineering and architectural services, account ing and auditing, and medical services. Other are intelligence service of wholesale traders or financial services. All of them are specialized in producing or selling information and advice (cf. Machlup 1962: 323f.). The ongoing chapter will concentrate on the financial services be cause it might be still one of the most important and developed industry for the use of AI within the information services nowadays. The primary goal is to look at the market and potential of AI within this field guided by the basic hypotheses of Analyse I and the following questions. How does AI revolutionize this specific sector? What are the supposed conse quences of an AI adaption within this industry? The market for AI in financial services is expected to grow from 1.3 bil lion dollars in 2017 to 7.4 billion dollars in 2022, at a Compound Annual Growth Rate of 40 percent, according to research and markets. A reason might be that the financial services industry is facing both numerous op portunities to innovate and challenges ahead that will determine the fu ture landscape of the industry post-disruption (cf. Fraser 2017). Many applications of AI within this field already exist. It has been creeping into financial services under a variety of names, assisted in no small part by related technologies such as digitalization, interactive voice response and image recognition or data mining for personal identity validation (cf. Narrativ Science 2017: 5). On the one hand, the adoption of AI use cases has been driven by both supply factors, such as technological advances and the availability of financial sector data and infrastructure. On the other hand, by demand factors, such as profitability needs, competition with other firms, and the demands of financial regulation (cf. Financial Stability Board 2017: 5). Nowadays financial services companies have been working on creating real-life AI use cases and are exploring new opportunities with hopes that strong AI could both cut costs and boost revenues (cf. Donovan: 2017; Williams-Grut 2017). However, it is esti mated that it will revolutionize finance through the ability to combine structured and unstructured data and provide precise market analysis (cf. Baker 2018). All in all, AI seems to be supposed to optimize financial services as an extensive solution for existing challenges. In the following part, a few concrete ways how especially strong AI could transform the financial service industry in the next years should be analysed. The first aspect is the enhancing customer engagement. At a time when the financial services industry needs to become increasingly fo 41 cused on creating better customer experiences, the importance of highquality and personalized communications is very important nowadays. For example, by using deep neural networks, different companies have already developed a conversational AI platform for financial institutions that is like a more advanced „Siri for your finances“ . It supports the same natural language How that a customer possibly has with his per sonal banker (cf. Chowdhry 2018). The imagination of strong AI can also definitely help to tackle this mis sion of enhancing customer engagement, both automatically and at scale. This personalized communications or advice as enabled by AI would be extended in several ways. For example, the customers own robo-advisors in the wealth management space that provide automated, algorithmbased portfolio management advice without the help of a human coun terpart. In this case, AI also automates the so-called „Know Your Cus tomer Processes“ for the mentioned wealth management. Also in terms of investment and asset management. AI is already starting to optimize this process that takes hours, or that often results in a human error (cf. Narrative Science 2017: 7; Bhardwaj 2018; Fraser 2017, Bashforth 2018). Research shows that evidence-based algorithms predict the future more accurately than human forecasters. “Artifcial intelligence can help people make faster, better, and cheaper decisions. But you have to be willing to collaborate with the machine, and notjust treat it as either a servant or an overlord.” (Bhardwaj 2018). An advanced level of AI and human collaboration could help human ex perts to identify issues and save time. The market size of robo-advisors, for example, seems to increase productivity as well and could eventually approach the so-called assets under management of the entire asset man agement industry in time. Similarly, AI-enabled personal finance intelli gence applications are helping consumers to manage their finances, ana lyze their spending, automate tax form filing, or making financial rec ommendations with a business model not predicated to generating fees from investments. This sketched automation of traditional retail func tions is made possible mainly through the creation of smart assistants and voice recognition technology (cf. Bashforth 2018; Fraser 2017). It also improves the targeting of messages at all and could end up with providing better customer support through AI and computer-based cus tomer services (cf. Technative 2017). In the field of customer support, especially the current state of the art of conversational AI for virtual as 42 sistance like chatbots helps companies to achieve more productivity and customers to transact or to solve problems faster. So a further use case of AI in financial services is to help custom ers in the field of decision support also by reducing essential risks. Per sonal assistance and robo-advisory services can help the consumer to spend and invest more wisely by giving augmented recommendations (cf. Narrativ Science 2017: 9f.; Bashforth 2018; Technative 2017; Bhardwaj 2018; Javelosa/ Houser 2017). But it can help the companies in the same way to minimize their own risks. For example in the fields of credit scoring the decisionmakers could use AI for robust credit lending applications to achieve faster, more accurate risk assessment, using ma chine intelligence to factor in the character and capacity of applicants (cf. Sigmoidal 2018; Financial Stability Board 2017: llff). Another famous use case of strong AI is being used in automated fraud detection and reduction, as well as anti-money laundering and anti terrorist financing compliance monitoring. “In most cases, the daily transaction volume is far too high for hu mans to manually review each transaction. Instead, AI is used to create systems that learn what types o f transactions are fraudulent/1 (Narula 2018). To reduce the fraudulent transactions is a huge benefit. But these fraud detection technologies are certainly not new. Regarding strong AI the innovative aspect is that AI and machine learning can mimic the associa tive memory of the human brain to identify likely fraud with an infinitely larger data set. Especially deep learning could further reduce fraud as it can take in thousands of variables versus a few dozens (cf. Fraser 2017: 4f.; Bashforth 2018). It is important because the most significant chal lenges in the financial and banking field are the complexity involved in identifying fraud. By leveraging better AI platforms and solutions, com panies could automate the analysis of massive amounts of data, which mean to fraudulent actors and other threats can be recognized and dealt with faster (cf. Bhardwaj 2018). They could also be developed into a predictive cybersecurity monitoring and response systems (cf. Bashforth 2018). All in all the client experience within the financial services can be fundamentally transformed through the power of strong AI. Morrissey, a technology strategist of Microsoft, emphasize that the future is an intel ligent bank for example, where technology can predict seamlessly what 43 the client needs next and empower them. Also, the next generation will demand new levels of personalization and frictionless experiences (cf. Morrissey 2017). In mid-term these improvements could end up in using much more bots then human employees. For example, to improve the targeting of messages or to provide better customer support and comput er-based customer services, AI can be seen as a job destroyer in the fi nancial services. It may be concluded that profound changes arising from strong AI have to be expected for the entire industry, even only a sub segment has been investigated here. 5.3. Information Machines Machlup further subdivides the sector Information Machines. He distin guishes between Information Machines for Knowledge Industries, In struments for Measurement, Observation, and Control, Office Infor mation Machines and Electronic Computers. However, the chapter will concentrate on the sub-segment Electronic Computers, here the case of smartphones. This for a simple reason: Smartphones already a part of our daily lives (cf. KeepCoding 2017). The internet connectivity of smartphones in combination with connectiv ity capabilities of other devices makes their importance evident, at least from a theoretical perspective. The machine smartphone communicates with other machines to obtain, process and sends data, which in turn is transformed again by other devices. Knowledge workers use smartphones intensively exactly for these purposes, for work tasks but also for private matters and thus turned into highly relevant tools: In this research smartphone can be summarized as a device which helps the knowledge worker to get more information, which influences his deci sions and consequential actions, even without the presence of strong AI. In the future, strong AI will also be able to find its way into the smartphone industry, which is one of the largest manufacturing indus tries today. However, to what extent does strong AI is going to change the smartphone industry and what might be the significant changes in short-term? Continuous efforts by companies such as Apple, Google, Huawei, Sam sung, and others gave birth to smartphone AI technology. Current usecases of machine learning and AI in smartphones are features such as voice-to-text translation, existing AI agents like Apple’s Siri, Google Assistant or common applications like Apple or Google Maps. These features and many more will expand across the industry, complemented 44 by embedded AI in all parts of mobile devices from cameras, to audio, to machine (cf. Fuertes 2017; Narula 2018). Emerging tech based on AI now enables smartphones to learn and recognize their users. In general, many predictions say that AI will trans form phones to their next iteration, because unusually deep learning al gorithms or neural networks, which recognizes sensory patterns as they happen. It is the reason image recognition, speech transcription, and translation have become more accurate. Upcoming capabilities in the field of strong AI will bring a further shift in technology and could fun damentally change the smartphone in the future. For example, Huawei's consumer head of software and intelligence engineering said (cf. Reichert 2017; Fuertes 2017; Morrissey 2017; Coughlin 2017): “I f you look at the whole ecosystem, the AI will fundamentally change the phone from the smartphone to the intelligent phone“ (Reichert 2017). Considering today's so-called AI applications in smartphones, at the core they are using machine learning and in detail a convolutional neural network, which is a type of weak AI and still many steps away from strong AI. The smartphones nowadays learn to recognize patterns from a significant amount of raw data, which usually takes place on massive server farms because it requires an immense amount of data. A machine learning application can then apply the patterns trained or learned in the first step to more data. For a smartphone, the latter step is most relevant. A trained neural network is already used for example to recognize ob jects or faces in an image (cf. Herget 2018). As mentioned above, cur rent devices that run AI algorithms depend on servers in the cloud, be cause the device itself lacks the horsepower to run AI algorithms. Using these servers limits how information is processed and is only working with online connectivity by sending data back and forth. A so-called on-device AI unit could change that process and increase the computing power of the device itself. It describes a processor dedicated to machine learning for mobile phones and other smart-home devices. This might be the most significant changes in the global smartphone market soon by climbing the way to strong AI. AI smartphones could be standard in the future by putting this chip in every single mobile device. It seems very attractive because could speed things up, cutting the lag inherent in sending information back and forth. It will allow hardware to run offline, which pleases privacy advocates, who are comforted by the idea of data remaining on the device (cf. Rahman 2018; Condliffe 2018). 45 Instead of always sharing data with a company server, AI can analyze the user's data on-device, which keeps it personal and under the user's control. In addition to this possibly better data protection, it could also improve the battery performance of the devices, as a consequence of data being processed and stored locally (cf. Lomas 2018; Coughlin 2017). “I t’s also predicting that, by 2022, a full 80 percent o f smartphones shipped will have on-device AI capabilities, up from just 10 per cent in 2017.“ (Lomas 2018). A few smartphones already have their on-device AI or also called the neural processing unit, which according to manufacturers is meant to make smartphones smart. In modem smartphones, it mainly includes functions related to image recognition, photography or processing (cf. Herget 2018). Primarily these upcoming AI chips can digest massive data sets based on the user's habits, daily patterns, and past behaviors. Lhis current type of AI can retrieve supporting information from mobile apps, fitness track ers, digital watches, or even browsing history. Lhe goal is to make pre dictions about what the user might do next. In the future, a standard ondevice AI for smartphones coming closer to strong AI and could make it possible, as mentioned above, that all analysis will be able to take place on a device without an internet connection, which is a revolutionary thought (cf. Coughlin 2017). Furthermore, this on device AI will change two key aspects of the smartphone shortly. On the one hand usermachine interaction, and on the other hand context-personalized open ness. Lhe first aspect will improve efficiencies between the user and his phone across text, image, voice, video, and sensors, while the second will actively provide services and aggregated information across apps, content, third-party features, and essential features (cf. Reichert 2017). It could be imaginable that strong AI might be the highest level of smartphones for everything in the daily life of the future. However, there are many other benefits of AI in smartphones. For ex ample, emotion sensing systems (emotion recognition) and affective computing will lead to an increase in personal assistance, which means better context and enhanced service experience. It will allow smartphones to detect, analyze, process and respond to his user's emo tions and moods. It could also understand the user's interests and tastes, even prioritizing notifications, more accurately. An apparent product of smartphone AI also enables devices to give predictions on how we can improve our lives. By that, a health app could scan the user's body, pull readings from phone sensors, and determine if anything is unbalanced. In 46 that case, users could be notified immediately. A few scholars estimate that soon the smartphone might be able to detect precursors for illnesses such as cardiovascular diseases, Parkinson’s or dementia. Even car man ufacturers could use a smartphones front camera to understand the phys ical conditions of drivers or gauge fatigue levels to increase safety. Other benefits are for business people, who are always multitasking, an im proved Al phone can de-clutter their calendar, schedule their conference calls, even record and transcribe notes from a presentation. It could also boost battery life, charge faster or increase storage space. Unless con sumer spending on mobile phones and apps slows down expect to see, these features rolled out in the future. Continuous training and deep learning on smartphones will also improve the accuracy of naturallanguage understanding (speech recognition), while better understanding the user’s specific intentions. The last promising application for Al might be augmented reality, where digital effects provide an additional visual layer on top of the camera or captured image. All in all the most noticeable changes Al will bring are processing speed, efficiency and new paths of opportunities (cf. Coughlin 2017; Lomas 2018; Fuertes 2017). Especially regarding the characteristics of strong Al in chapter two, its use-cases in smartphones are still very young but poised to have a transformational impact on the mobile device of the future. It will cre ate a personalized, more and more user-friendly and beneficial relation ship between the smartphone and its user. The smartphone is on the way of being a perfect personal assistant for everyday use of the consumer even for information worker because strong Al will bring smartphones to a much higher level in the future. 5.4. Media ofCommunication The preceding analysis has shown the potential for signification effects of strong Al on industry segments. It may be assumed that this will also be the case for the segment Media of Communication. Changes in ICT will profoundly impact today's media structures, the media industry will change too. Substantial amounts of goods produced by the media indus try can be digitalized. The internet and mobile devices are an ideal means to distribute digitilized goods. Consequently, the media commu nication and distribution of content shifts from traditional media like pa pers, magazines and books and outlet selling these goods to internetenabled devices such as the smartphone (cf. Dolata/ Schrape 2012: 7). 47 One goal of this chapter is to explore current and near-future use cases of AI in Media of Communication. Using the approach again to dwell deeper into sub-segments of an industry, the chapter looks at the sub segments advertising and broadcasting. Other communications based on smartphone apps, social media, email, online chat rooms or communica tion with chatbots are not being investigated. So all the applications, cit ed in the following chapter will by no means be exhaustive or definitive for the Media and communication as a whole. However, the selected sub-segment may still help to find insights into the validity of the two stated hypotheses: the impact of strong AI on job availability and knowledge workers in the Media of Communication Industry. Before starting the analysis, a closer definition of the industry shall be applied. The Media of Communication cover as well as oral communica tion media as written communication media. In detail oral communica tion describes the process of communication in which messages or in formation is exchanged or communicated within sender and receiver through word of mouth. It can be sub-divided into speaking and listen ing. Written communication includes messages or information, which is exchanged or communicated within sender and receiver through written form, so writing or reading (cf. The Business Communication 2018). The first analyzes object will be the advertising sector. At first glance, it can be seen that digital advertising and media might be very suitable playgrounds for strong AI. The already used applications of AI in advertising and its connected media are as varied as they are numer ous, ranging from ad targeting, content curation, and creation, or dynam ic pricing, through fraud prevention, predictive customer service and product recommendation, to programmatic buying, sales forecasting, and web or app personalization. Through the combination of different tech nologies based on AI, consumer experience could be brought to the next level. The fact is that humans are simply not able to analyze and interpret all available data that can be gleaned from consumer habits. Sure, they can spot patterns and act accordingly, but this is incredibly inefficient when compared to what the current level of AI can already do and what especially strong AI could do in the future. Strong AI and machine learn ing algorithms can leverage billions of data points to make predictions for future campaigns, and then use real-time information to optimize throughout. The following scenario could picture one promising imple mentation of AI in advertising, based on the assumption that a fashion style of person gives information about its buying behavior: A customer enters a store where an image recognition system identifies her fashion 48 style for example. The profile is then fed into a second system that uses the preferences of consumers with similar styles to recommend specific products or colors, or else to deliver the most suitable content via in store displays (cf. Maugain 2017, Fritschle 2017). In this sense, strong AI in advertising could mean new uncovered opportunities for brands, customer enhancement by understanding their behavior in marketing or an appropriate targeting of advertising in a sense, bringing up the right information, at the right time, through the right channel. The second analyze object is broadcasting. Compared to other in dustries, there seem to be fewer articles that focus on AIs application to broadcast, or how it delivers unquestionable and immediately realizable benefits. However, it is not easy to determine which segments of the broadcast or even its connected media industry get the most significant impact from developments in AI. Based on the qualitative analysis for this research, it can be argued that broadcast can take advantage of strong AI by providing new opportunities for helping to enhance metada ta and provide closed captioning. Ethan Dreilinger (IBM Watson, solu tions engineer) said one of its main early beachheads is in diving into broadcasters extensive archives and cleaning up its metadata to improve retrievability, along with real-time closed captioning services. Further more, there might also be the promise image recognition for audience analysis as a potential mainstream application. On content, recognizing images from live or recorded videos could lead to metadata tags or de scriptive information of a movie or a program beyond its usual essential attributes such as genre, language or actors. Due to the potential of strong AI, it is also starting to learn the roles of an editor or creative di rector for television commercials and may soon help suss out fake news. Concerning more general use cases for AI in content creation and deliv ery, the two most viewer-centric applications would be on content dis covery and content personalization. Its primary goal is to deliver person alized and targeted experiences to multiscreen audiences to keep them coming back. Furthermore the same applies to ad targeting because ad vertisements can also be tailored to the audience - based on their de mographics or preferences. All the applications cited in this chapter are by no means exhaustive or definitive. However, it seems evident, that we will see growing adoption of AI-powered solutions in different aspects of broadcasting. For exam ple from enhancing metadata, providing closed captioning or content creation to management, delivery, or consumption (cf. Fronda 2017, Depp 2018; Lant 2017). 49 Based on the elaboration on the Media of Communication, it may be concluded - as for the other investigated industries - that profound changes arising from strong AI have to be expected for the entire indus try, even only a sub-segment has been investigated here. 5.5. Research and Development Machlup describes research and development as a sub-segment, which combines two knowledge-producing activities. On the one hand, new knowledge about how things are or how things could be made is origi nated in the mind of the researcher, discoverer, inventor, or developer. On the other hand, this knowledge is also produced in the minds of oth ers. In the first part of the chapter research and development should be defined more precisely. Regarding research, it has become customary to distinguish between basic and applied research work. The general idea behind the distinction between basic and applied research is that basic research creates basic knowledge, on which practical, applicable knowledge may rest but which itself is too general, too broad or too deep, to have direct applications. In contrast, applied research creates directly applicable knowledge. So the main difference is while in basic research the investigator looks for general laws, with no regard to practi cal use, in applied research he looks for results which promise to be of ultimate use in practice. For example, the better understanding of the physical or organic world is the goal of the basic research, better prod ucts or better ways of making them are the goals of the applied research. It sounds like a rather clear contrast between the two types of work, but in reality, the borderline cases are very numerous. Even more difficult than the distinction between basic and applied re search is the separation between research and development. Not only the question of where development starts is hard to answer, also the question of where it ends is equally troublesome (cf. Machlup 1962: 145 ff.). So there might be many complex and intertwined distinctions, not feasible to handle in this chapter. Against that backdrop, there should be no spe cific focus on research or development. All in all, there are very fewer starting points for strong AI in both fields anyway. One reason might be that as well research as development is too abstract and complicated for being done by AI-driven resources. Human skills such as multidimensional thinking, recognizing structures or creativity are essential. Regarding the current state of the art strong AI seems very far away to disrupt the fields of research and development. 50

Chapter Preview

References

Abstract

The ongoing and seemingly unstoppable digital transformation brings forth new options, opportunities but also challenges to individuals, organizations, companies and societies alike. Governments are alarmed, realizing the potential consequences on the workforce, while also being apparently helpless against uncontrollable and powerful digital players such as Google or Facebook. As Henry Wittke shows, recent breakthroughs in the field of machine learning increase the potential of Artificial Intelligence to disrupt the world’s largest industries. Wittke attempts to provide a basic framework of what constitutes AI as well as to assess its impact on the Information Economy. What happens in case of rising mass unemployment or social inequality? What will be the effect on labor as a value system for today’s societies? Could the entire notion of capitalism be questioned in the wake of AI? The book aims to draw conclusions and give recommendations to policymakers.