Will make IBM Watson data scientists obsolete

IoT - Consumers and the Internet of Things


Internet of Things, Artificial Intelligence, IoT, AI, Augmented Intelligence, human-computer symbiosis, autonomous systems

download the article as a pdf

Machine alone or man plus machine?
Recent developments in sensor and network technologies have led to a significant increase in the spread of applications for the Internet of Things (IoT), e. B. so-called wearables. These are based on artificial intelligence (AI) technologies such as machine learning and vision, deep learning, language processing or big data analytics. The advances in this have in turn led to the development of autonomous systems in the field of smart homes, smart cities and smart grids or driverless, autonomous driving. Will such intelligent IoT systems soon make human labor redundant? Will people still play a role in the working world of the future? Will AI systems completely replace human beings, or will they rather expand human intelligence and enable us to solve the most pressing problems of our time better than we ever dreamed of? I assume that an effective symbiosis of people and machines - in other words "Augmented Intelligence" - has the potential to successfully solve some current challenges and that it will work better than pure AI solutions, at least in the foreseeable future.

Artificial Intelligence or Augmented Intelligence
Augmented intelligence is understood to mean systems that extend human intelligence, while AI usually refers to computer applications that completely replace humans. The hype surrounding the potential of the IoT mostly affects the self-controlling, autonomous AI systems that work using big data analytics and obtain their data from IoT applications. The fact that IoT applications can collect, aggregate and analyze the enormous amounts of self-generated data is a basic requirement for harnessing the potential of the IoT - both for pure AI solutions and for those that support human intelligence.

Machine learning, language processing, data analytics and other AI applications deliver better and better results even without human intervention. This is different in areas such as design, creative advertising, personal sales, employee selection and mentoring, in strategic decisions or the treatment of diseases. These represent important fields of application in which human assessments and intuition are still important, and a human-machine symbiosis usually brings better performance. Human intelligence is still needed when it comes to decisions relating to personal data about employees, customers or partners and their preferences, behavior, habits, emotions and the like. If only ambiguous, vague or incomplete information is available, humans are fundamentally superior to machines. A lack of emotional intelligence and questionable judgment are still the biggest obstacles to the use of AI. Augmented intelligence integrates the unique human capabilities that AI cannot replicate, and many visionaries like Bill Gates, Stephen Hawking and Elon Musk have legitimate doubts about artificial intelligence. Far-reaching IoT problems can often be solved neither by pure computer solutions nor by humans alone. This is why there is great potential for IoT applications that rely on augmented intelligence. Machine learning approaches and human intelligence should be coupled in such a way that people ultimately remain the highest level of control.

The potential of human-computer symbiosis in the IoT
Human-computer symbiosis is understood to mean any type of collaborative interaction between humans and computers. In the IoT context, such a symbiosis arises when data is collected via the IoT with which AI applications make standardized calculations. These are based on criteria that people have set and are intended to generate insights for evaluations and decisions. The basic assumption of the human-computer symbiosis is that the strengths and problem-solving abilities of humans and computers complement one another. Augmented Intelligence arises when we optimize the computational performance of computers - combined with the cognitive abilities, intuition and common sense of people.

The concept of augmented intelligence is not all that new and related to research on human-computer interaction (HCI) that has existed for some time. Interestingly, HCI researchers have always warned that AI over-simplifies things by viewing people as perfectly rational machines. The HCI approach, on the other hand, aims to improve the symbiosis between people as emotional and interpretive units and computers in order to increase human performance. Many augmented intelligence approaches use crowdsourcing strategies and gamification that correspond to the spirit of the HCI tradition. You develop augmented intelligence designs for IoT devices to improve the interaction between people and IoT technology.

Box: Commercial applications of augmented intelligence in the field of IoT

All high-tech giants are promoting the development of augmented intelligence. Commercially available IoT platforms such as Amazon AWS, IBM Watson or Microsoft Azure are making rapid progress, and many of these applications correspond more to an augmented intelligence mindset than pure AI logic. Google's approach to its search engine design is also comparable to the tradition of augmented intelligence. The platforms are constantly expanding their range of services and their distribution by continuously integrating new technologies and services.

One of the best known and best developed platforms is Watson from IBM. Named after IBM's first CEO Watson, it was developed by IBM's research department in 2007 as part of a project that resulted in a question-and-answer tool for taking part in a quiz show. The tool actually won the show in 2011. It was then further expanded and commercialized in 2014. Today, Watson offers "cognitive computing" and supports people in reasoning and making decisions through speech and image recognition. While classic computers have to be programmed, Watson understands the world in a similar way to us humans: through learning, interpretation and experience-based improvements.

From health or education to finance to transportation and energy, Watson is trained by leading experts in their respective fields. The system masters seven languages ​​as well as the special characteristics and technical terms of a wide variety of industries and accesses well-founded, specialized pools of knowledge to enable people to make faster and more well-founded decisions. "Cognitive systems are especially helpful in the heavy work - putting data together, analyzing information and generating relevant responses from it, so that users can make more secure and more effective business decisions," says Vincent Thomas, Client Engagement Leader, IBM Watson.

This is how the Internet of Things becomes more human
The IoT, whose multi-networked devices exchange information and make decisions without human intervention, leads to justified concerns among managers: Can you really trust that AI can get by without human control? Organizations are faced with numerous risks that AI causes in IoT devices, such as data protection problems, mechanistic decision-making processes or loss of control. Therefore, managers should carefully examine for which task, in which way and to what extent IoT applications should be used. The following guidelines can help managers to develop optimal solutions for the tension between Augmented Intelligence versus AI in the context of the IoT.

  • Analyze the tradeoffs between full automation and human control
    Managers can rely on fully automated AI solutions as well as follow the augmented intelligence logic. Decision criteria are the expected performance, the costs and the risks of fully autonomous solutions. IoT services that could be carefully fully automated, for example, are automated production, predictive maintenance and security aspects. On the other hand, in applications that are more human-related, such as smart retailing, human controls could be built in. Beacon technologies and gaze trackers could, for example, optimize the placement of goods, while at the same time salespeople with mobile IoT devices call up personalized information and thus make IoT solutions even more powerful.
  • Design the IoT user interfaces carefully
    Augmented intelligence applications are becoming more and more common in many IoT areas such as cybersecurity, counter-terrorism, health care and space travel. The interaction between people and such IoT applications does not become more productive all the time. IoT developers should focus more on human-machine interactions and user interfaces. By following the HCI logic, you can make the resulting IoT systems more efficient and effective and simplify human control.
  • Use synergies between AI and augmented intelligence
    Managers should discuss more about possible synergies between augmented intelligence and AI in the IoT and how one can effectively couple human cognitive abilities with the computing power of computers. This can result in symbiotic human-computer applications in a wide variety of areas such as health care, fin-tech, smart cities or smart grids.

Superfluous or empowered: The long-term prospects for the Internet of Things
In the distant future, machines alone may dominate decision-making processes in many applications. Until then, however, a fairly long transition period can be expected, in which great intellectual advances will be achieved through close cooperation between humans and computers. IoT is an ideal application area to solve problems through the integration of human or machine intelligence. In addition, our society needs time to adapt to the major social, economic, behavioral changes and ethical issues that AI and augmented intelligence bring with them. Both areas concern the future of work itself, the productivity of companies, the blurring of industry boundaries and other legal and socio-political tasks of governments. In the near future, suitable IoT designs should therefore retain a reasonable level of human control and control and give humanity the chance to become familiar with the delegation of control to machines.

Author / s

Paul A. Pavlou, Milton F. Stauffer Professor and Co-Director, Data Science Institute, Fox School of Business, Temple University, Philadelphia, PA, USA, [email protected]


Licklider, J. C. (1960): “Man-Computer Symbiosis”. IRE Transactions on Human Factors in Electronics, (1), 4-11.

Meyer, G .; Adomavicius, G .; Johnson, P. E .; Elidrisi, M .; Rush, W. A .; Sperl-Hillen, J. M. and O'Connor, P. J. (2014): “A Machine Learning Approach to Improving Dynamic Decision Making”, Information Systems Research, Vol. 25 (2), 239-263.

Winograd, T. (2006): “Shifting viewpoints: Artificial intelligence and Human-Computer Interaction”, Artificial Intelligence, Vol. 170 (18), 1256-1258.

https: //www.ibm.com/blogs/watson/2017/01/top-3-benefits-cognitive-comput ... (January 11, 2017 | Written by: Trips Reddy) (quote)

https: //www.computerworld.com/article/3129232/big-data/watsons-the-name -...

  • Nuremberg Institute for Market Decisions e.V.
    Founder and anchor shareholder of GfK SE
  • NIM
© 2018 Nuremberg Institute for Market Decisions e.V.
All rights reserved