That’s NOT Artificial Intelligence

That's not Artificial Intelligence

Artificial Intelligence has been the talk of the town in both the business as well as the scientific field. It is seen as the next step in our revolution and an invention that should be applied within our society in many different areas. Artificial Intelligence (AI) sounds fancy and valuable when it comes to the development of products and services for organizations. It is becoming a unique selling point. However, do not be fooled, what is presented as AI is not always really AI. There are many examples in which AI is interpreted in the wrong way. For this reason, it is wrongly presented and misused by the press, in business communities, and in everyday use by the normal population.

In this article, we want to discuss what AI really is and how it is currently misinterpreted and misused in different areas. We will also go deeper into the reason why we have come to this state and which parties play a role in this. The consequences of the current situation for AI will be discussed and some ideas for possible solutions will be explored. By the end of this article, you will understand what AI really is and what is currently wrong within our society when it comes to the use of AI.

Definition of AI

The Artificial Intelligence field is mainly about defining intelligence and finding ways to build systems that act in an intelligent way. Therefore, the definition of Artificial Intelligence itself has changed over time by changing what we perceive as the definition of Intelligence. There is a root cause for such an evolving definition for a fundamental concept like intelligence. Humans tend to identify the abilities that distinguish them from other creatures as intelligence. As machines become increasingly capable, tasks considered to require “intelligence” are often removed from the definition of AI because they are no longer a distinguishing factor for the cognitive abilities of humans. This is called the AI effect which makes it intrinsically hard to put a definition for the field of Artificial Intelligence. As Tesler’s Theorem says “AI is whatever hasn’t been done yet.”

AI is whatever hasn’t been done yet.

Tesler’s Theorem

With that in mind, the AI field can be defined by the main pillars of intelligence which are the challenges the research in the field is trying to overcome and the main approaches of achieving those goals and making a system characterized by intelligence. By the time of writing, the research and development in Artificial Intelligence include challenges like Reasoning, Knowledge representation, Planning, Learning, Natural language processing, Perception, Motion and manipulation, Social intelligence, and General intelligence. Approaches include statistical methods, computational intelligence, symbolic AI, and cybernetics.

AI is being misused

Overall, there are three kinds of groups of people when it comes to their knowledge of AI. The first group is the innocent people, they don’t really know what AI is and are influenced by others who claim that something is AI or know what AI is. The second group is the people who are using AI incorrectly. This group could be misinformed, but could also be the group (or organizations) who are misusing AI for their own benefit. The third group is the people who really know what AI is and use it correctly. The first and second groups are the ones that cause problems and will influence the AI field the most, in a negative way.

We have come to the conclusion that there are two ways of misuse. The first way is that companies choose to introduce themselves as AI companies or that their products include AI while they don’t really include any kind of AI. The second way is that people are making limited use of the AI field because people don’t know that there is more to AI than one of the pillars. There are also two reasons why this misuse happens. The first reason is that people could deliberately misuse the term AI for their own benefit. The second reason is the misinterpretation of AI, which means that people understand it incorrectly or that they are (accidentally) misinformed by others.

Organizations may have the good intention to make an AI application but in reality, this is not the case. Organizations have to be careful when it comes to AI and should be aware of the pitfalls that can slow, limit or ruin their AI initiatives. One of these phenomena is AI washing. AI washing means that an organization positions a product that involves AI when this is not the case. They might just make use of basic forms of data analysis which they think makes something more intelligent. 

We have noticed that companies use AI in their sales pitches for clients but also in campaigns for the mass public. A report by MCC Ventures was published in 2019 that confirmed that 40% of the European companies that are classified as an “AI startup” does not exploit the field of AI in any of their business. Out of the 2830 startups that they investigated, only 1580 fit the description for an AI company. Companies classifying themselves as AI companies benefit from this when it comes to investments. An AI startup can attract 15% to 50% more funding as indicated in the same report.

Not only startups classify themselves as AI companies. Big organizations use this term as a selling point as well. In 2018 Android Authority brought to light that some tech companies are incorrectly claiming that their products have AI features. Asus, for example, positioned some of the features in their new smartphones as “AI” while they did not make use of any form of AI. It was said that there exists “AI Charging”, the phone monitors your charging habits to extend the battery’s lifetime. In reality, this is just no more than a maximum overnight charging capacity. They also mentioned “AI Scene Detection”. The idea behind this is that it will automatically select between different scene types. Again, a feature that already exists when you set the camera on AUTO mode. A result of this misuse is that it will confuse customers.

AI startups attract 15% to 50% more funding

As mentioned, there is also a party that as a result of misinterpretation limits itself in its use of AI. Companies or people are calling themselves AI experts, but in reality, they use one pillar of AI, for example, Machine Learning (ML), and treats it as if this is AI. AI and machine learning are being used interchangeably, which still causes confusion for the mass public and the media. Over time, the difference between AI and ML becomes unrecognizable. Then they even start forgetting to mention that Machine Learning is just a part of AI. Companies should also not call themselves Machine learning companies, the business should not define themselves by their tools. Machine learning can only solve particular problems, while when you look also at the other pillars of AI you could solve many more problems. 

As you can see there are many examples when it comes to companies and their use and interpretation of AI. Within the mass public, there is a lack of knowledge about what AI really is especially for those who don’t come by AI often. In 2018 Elsevier published an article in which different professors discussed the misconceptions of AI. When it comes to the mass public the biggest misconception is that people think that we are close to AI. We should not overestimate the standards. Yes, AI can do some things, such as classification and speech and facial recognition. However, machines that can reason about the world like humans cannot exist yet. For this reason, it is also important that the mass public is well informed about AI in general as well. 

Motivations and Benefits

The next question that we should ask is; how and why did we get to this state? We already mentioned the issues that have developed around the understanding of AI. It can’t go unnoticed that there are certain motivations for the misuse of AI in general that benefits certain parties. Furthermore, this misuse does not only benefit certain parties but can also benefit the whole AI industry!

The first motivation to misuse AI is that “AI sells”. AI is a trending topic in the press and many organizations want to jump on this AI train and offer their clients the newest tools and techniques. Products that include “AI” are seen as innovative and progressive and are very attractive for customers. Including some “AI’’ in the sales pitch benefits organizations when it comes to attracting new clients and selling more products. 

Furthermore, salespeople like to include “AI” in their sales pitch because people won’t question it. Not many people are familiar with what AI really is and how it works. This lack of knowledge from the client/customer side gives the salesperson an advantage. If you don’t understand it you will easier accept and agree with the information that is given to you.

A second motivation is that AI sounds ‘’prestigious’’. Working with the latest technology or being able to offer it gives someone or an organization a unique position. Having experts in a certain field makes you look innovative and a trendsetter. Other people or organizations will look up to you. It also gives you as a company the opportunity to knock out your competitors. 

We have seen that the current situation around AI brings not much good so far. However, It turns out that there are at least two benefits. First, artificial intelligence as a field gets a lot of exposure. The press pays attention to it and spreads the message to the mass public. Organizations can no longer ignore it and many have already started using it. More people are getting familiar with AI in general and many conversations are started. Benefits, risks, and opportunities for our society are being discussed. What kind of role AI is getting or should get in our society is becoming an important topic when it comes to future developments within technology and our society. 

Second, the increasing exposure to AI leads also to more investment in the AI industry. This means that investors are spending money on AI start-ups and that organizations reserve budget for AI employees or tools that they need to purchase. Furthermore, more research is being done in the AI field by scientists, which leads to new discoveries. These new discoveries are important for both improving the current as well as developing new AI tools, products, and techniques. Investing in AI in general opens new doors for the prosperity of our society.

Bad Consequences

The misuse of the term Artificial Intelligence has very bad consequences for the future of AI. Using AI to market and sell products that have nothing to do with AI hurts the credibility of AI. Those companies or startups claiming they are using AI while they are not, attracts unfair investment which should have been put into the real research and development on AI. Misinterpreting AI as only being one of its pillars, like Machine Learning, results in an underdeveloped and limited future for AI. Most of the focus is going into developing and investing in specific areas of AI ignoring the bigger picture. This limited focus will trap the industry and the field of AI into a vicious cycle putting a ceiling to what could achieve using AI.

This short-sighted view of AI also limits the innovation of the workforce because it is only aware of and uses one form of AI. We can see many people rushing to learn Machine Learning and many job openings for Machine Learning which leads to a workforce only aware of machine learning techniques and ignorant of the other parts of AI.

The limited focus will trap the industry and the field of AI into a vicious cycle

This also happens on the organizational level. When organizations only recognize one pillar of intelligence as AI and only choose to dive into this technique they will miss out. Their investment in AI specialists will only go into investing in people with a certain skill which will result in insufficient skill investment. The organization will live in an illusion that machine learning can be used to solve all the problems and end up trapped with a tool that is only feasible for a set of tasks or they will not reach the highest potential of AI because they are limiting themselves to one form of it.

We can already see this limited focus on specific subfields of AI in numbers from the AI Index. The AI Index is a yearly report produced by Stanford University. The report provides unbiased, rigorous, and comprehensive data about the current state of the AI field. In the 2019 report, we can clearly see how far the industry is driven towards specific areas of AI which witnessed serious advancements in recent years. On the research side, Machine Learning has become the largest category of AI papers on arXiv in 2019. We can see the same trend in job posts. About 77.6% of the job posts related to AI in the US are for Machine Learning, Neural networks, NLP, Robotics, and Visual Image Recognition with Machine Learning having 40.8% of the AI job posts. The same observation can be replicated in education and web search keywords frequency.

Possible solutions

We have to admit the unfortunate fact about the current situation of the AI industry. That would be the first step on a road full of raising the awareness of various parties about the real and bigger picture of AI. While many parties need to be involved to achieve a better future for AI, we believe that one of the powerful tools that have a great hand in rectifying the current situation is Consumer Protection. Consumer Protection is aimed towards safeguarding buyers of goods and services, and the public, against unfair practices in the marketplace to gain an advantage over competitors or to mislead consumers. We believe that Consumer Protection organizations and agencies should pay more attention to the usage of the term AI and start working on laws to regulate it. This will definitely help limit the spread of AI washing.

Another angle on which we can attack the current situation is related to how and when AI is chosen to be incorporated in a product or a service. We regularly see startups classify their products as AI even before developing the product. AI is not always the best solution to the problem and it is inappropriate to claim that a product is or uses AI beforehand. Starting the product development from such a biased perspective results in products that are marginally touching AI because the organization wants to label them as AI, or even worse using AI for something that doesn’t require AI. AI needs to be treated as a tool to be used when it makes sense, not a predetermined decision that AI has to be involved in the product no matter what.

Conclusion

In conclusion, it has become clear that there are some things that we have to work on when it comes to defining the AI field. The misuse of the term is misleading the customers, the business, and the press by giving people the impression that something is AI when it is not AI or treat one pillar the whole AI. Even when this is giving AI a great amount of exposure and increases the number of investments going into the field, it has to stop. It is important that we aim towards a future where the AI term is well understood by all parties. In this way, we can prevent confusion, unfair investments, and the spread of incorrect interpretation of AI. Organizations should also explore more than one pillar of AI and broaden their AI knowledge and expertise. We should put out a call to action to different involved parties. We urge the press to use more complete interpretations of AI. We advise people to be more critical when they come across AI. The future of AI is on the line, we just started to explore it, let’s not ruin such a promising chance to make the next biggest leap in our society.


Leave a Reply

Your email address will not be published. Required fields are marked *