Computerwoche: How Autonomous Process Automation Simplifies Our Lives

René Büst, Director of Technology Research at Arago, is published with an opinion piece on current artificial intelligence technologies in one of Germany’s leading weekly IT-magazines Computerwoche. Throughout the article Büst argues that we should not be afraid of AI and have lower expectations about what the technology has to offer at the moment.

He is convinced that Elon Musk’s and Stephen Hawking’s current fear-mongering about mass unemployment would essentially misguide people. It creates a perception that AI research and development would have made quantum leaps recently. Yet, current technology would be at the beginning. In this way, these negative predictions would only create fear and opposition to AI. Büst believes that the challenge would rather be to inform citizens what AI actually means. A recent poll would have shown that 43% of Americans and 46% of UK citizens had no understanding what AI entails. At the same time, Büst argues that regulations on AI and ethics are relevant. In particular, the question of accident culpability regarding self-driving cars would be a central question on people’s minds.

Since we are currently living in an age of exponential technological advancement, a “will-never-happen”-attitude would be wrong. However, the most successful AI systems are only able to solve strategy games such as FreeCiv – as in the case of Arago’s HIRO. In addition, most AI-projects would be based on machine learning, a technology that helps to recognize patterns in large data pools and is able to make predictions based on this data. Such programs could only deliver reliable results if they have access to high-quality data. Machines always find some kind of pattern in large data pools. Therefore, it would still be essential to check the results on their accuracy and plausibility. Büst considers this a major shortcoming of machine learning, as this technology requires extensive amounts of data to come up with valuable results. The current aim of AI would not be to emulate a human brain but to deliver a system that acts like a human. In this context, Büst asks his readers to consider AIs as autonomous process automation which could add more to our daily routines.

Even though Amazon’s Alexa and Apple’s Siri are not able to hold a conversation, they are AI technologies based on machine learning algorithms with prediction models. Yet, neither Alexa nor Siri would be intelligent or self-learning. Their knowledge does not come from actual brains but from the cloud databases of Amazon and Apple. Besides Amazon’s ever-growing database, Alexa learns through so-called “Alexa Skills”- little applications that add extra questions and answers to the pool. With every additional skill, Alexa seems to become more intelligent. In Büst’s view, the most interesting aspect about Alexa would be the team of 5000 employees currently working on the technology. This would promise much progress in the future.

However, self-learning systems would already exist as well. For example, the iPhone is able – if connected to one’s car system via Bluetooth – to make predictions about estimated arrival times to a home location or frequently visited places. These results are all just based on one’s travel routines. In the same way, Google Now could make predictions about one’s searching behavior and functions as a personal assistant. Büst also names other examples of AI-related services that are already on the market for consumers: credit rating systems at banks, dating websites making predictions about dating probability as well as robots helping travelers with check-in and room service at airports or hotels.

In addition, he presents a couple of ideas how AIs as autonomous process automation could make our lives easier. At the office, Alexa and Siri could act as a personal watchdog; take calls from colleagues and schedule appointments autonomously. While travelling, they could independently schedule a taxi based on traffic and one’s personal schedule. In the kitchen, they could make suggestions about meals based on the contents of the fridge or give advice about a healthy diet.

Büst also presents two real-life examples of autonomous process automation in enterprises. With an AI-defined infrastructure in the IT-department, companies would be able to set up a self-learning and self-repairing system. Without human interaction, the AI can allocate IT resources to certain tasks – depending on their workload. In this way, it can react to issues within the infrastructure and can repair the system proactively. In the insurance industry, autonomous process automation is deployed to issue insurance contracts independently. After learning about the creation process of a contract, the system is able to draft the insurance policies and sends its results to an expert. This expert potentially amends the policy and hands it over to the client. The target would be to create individual insurance policies based on the personal circumstances of every client.

As a conclusion, Büst argues that the biggest threat to AI-research would be human impatience. He asks his readers to remain calm and to understand that AI research and products would only improve step by step. Otherwise, expectations on AI could not be fulfilled and we would head into an AI-winter. AI technologies should be understood as an approach to make our lives easier. Yet, the word “intelligent” should be used carefully and users should be careful about what kind of data they want to share with an AI. Once they are out of one’s personal control their processing could not be reversed.

Click here to read the article

Arago Redaktion 10. January 2018

Contact us

Arago Arago GmbH info@arago.co