Episode 1: “What the heck are we talking about?”
“I’m excited about the accelerating developments in technology. AI is transforming daily life and business operations in a rapid pace, news on progress fly in weekly and even daily. With that in mind, it is time to look at these trends and significant developments from a wider angle and separate them into individual fields.”
Steve Nitzschner, CEO & founder of Wildstyle Network GmbH, Chairman at “AI in automotive” 2019
While everybody seems to be talking about it, AI can certainly be a confusing topic: Definitions are still evolving. New businesses, industry segments and tasks are emerging constantly and on top of that, clever marketers not only keep creating fancy new terms for what is essentially the same thing, but have also started referring to simple and decades-old algorithms as AI with the intention of pushing companies or products. “In 40 percent of cases, we could find no mention of evidence of AI,” claims MMC head of research David Kelnar, editor of “The State of AI: Divergence 2019”[1].
This research analysed every European startup that self-identified as AI-based in 2018: Nowadays, new or existing services sell better if they are smart, automated, self-learning, well-trained and UX friendly.
Starting in 1951, when the first AI program – a chess game – emerged, consumers have come to expect software to be intelligent. Since then, consumers have become increasingly curious about the idea of “smart” software. All that marketers did in this case was to embrace this perception and make an excellent move by claiming new products to not only be smart, but supposedly even have built-in AI technology. This helped generate a constant demand for technological improvements and established a new industry standard.
For the experts, a true AI-based technology surpasses the stage of supervised learning.
The majority of companies working in this field know that “real AI” is starting at the point where the use of self-learning algorithms follows computer vision to an AGI Level [2]. From an average consumer perspective, products such as Alexa appear to be able to read people’s minds but how intelligent are these products really? Amazon eventually had to reveal that it is, in fact, employees who analyse thousands of records per day, using human decision making skills – not AI – to classify customer voice sequences in order to improve Alexa’s natural language processing (NLP) algorithm.[3] To understand the complexity of AI, it is important to determine where technology really stands at present. We are still in the era of supervised learning which is technically not classified as AI. Yet, learning algorithms such as „deep learning”, newcomers in the field of „Evolutionary Algorithms” and neural networks are evolving at a rapid pace. An increasing number of human abilities are converted into algorithms which help upgrade the complexity of products across all industries. While certain verticals may overlap, others remain unique to the individual case. Amazon, for example, might add emotion recognition as a so called “skill” to Alexa[4], while car manufacturers might work on a digital co-pilot instead. Hence, we see that VUX [5] and cloud-based data are used in both industries and are adaptable, whereas the algorithms for autonomous driving remain product and hardware specific.
Automation and artificial intelligence are not just a volatile trend as is the case with many new products or technologies. From a high-level perspective the evolution of AI can be divided into the following three tiers. From a high-level perspective the evolution of AI can be divided into the following three tiers: Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI) and Artificial Super Intelligence (ASI). Currently, AI technology is still being developed within the framework of tier 1 and used in Google Search, Alexa NLP and autonomous driving solutions. Additionally, the AI platform landscape offers tons of single services for different data processing or image recognition solutions.
These functionalities are complex and definitely worth the billions of dollars invested in research and testing. Considering the public hype, it is safe to say that the technology is currently still stuck in the stone age. At this time, “thinking” technology is still limited to its domain and has coders, data analysts and huge trained datasets as dependencies. Hence, AI functionalities are still rather basic and completely supervised.
Although the ANI tier still requires a lot of men and computing power paired with experience in order to reach its full potential, a few scientists and companies are already thinking ahead to necessary development on the AGI tier. At this stage technology would move beyond an algorithmic illusion and start mimicking human intelligence and behaviour. Based on unsupervised learning, the AI would grow across domains and be able to make decisions which would not be pre-trained. Scientists presume that the approach of “Reinforcement Learning” could turn out to be the holy grail regarding AGI. A breakthrough on this tier would fundamentally change the product and service landscape.The final tier is ASI which is largely dependent on the velocity of progress in the field of robotics. At this time, limitations are set only by the law of nature. Taking into account the already heated discussions about “dead-bots” which, at this stage, are still nothing but a working business model [6], imagine the ramification scientists across multiple fields will have to face in the future. It will be an extremely important and interesting phase during which important ethical and moral questions, especially concerning artificial humans, will have to be answered.
Staying on top of the developments in the field of AI is a difficult task if you don’t know how to interpret the flood of content streaming in on a daily basis from sources across all industries. Companies are investing heavily in marketing and are publishing a multitude of research studies – obviously with the goal of creating ever-growing revenue streams. The field of AI topics have reached an astonishing level of differentiation and our AI department at Wildstyle Network is constantly working to stay ahead of the curve with regards to industry updates.
Fact is that 82% of companies have produced a financial return on their investment which, for example, shows in the median returns of 17% on investments in cognitive technologies for companies across all industries. This year’s “Handelsblatt AI in automotive” conference, held on December 3rd and 4th, 2019 in Munich, will be presided over by Steve Nitzschner, CEO and founder of Wildstyle Network, giving him the opportunity to share his expertise in the field of artificial intelligence.
In collaboration with our AI department at Wildstyle Network, we will publish a series of weekly articles sharing our unique and up to date perspective on the latest trends regarding the complexities of our technological future. Every article will provide an overview and summary of select AI-related topics.
Editors:
Steve Nitzschner
Jens Franke
Manuela Krahl
Mandy Pearson
Alexandra Kloppe