Breaking down myths about AI – In these three things people have the wrong idea
Artificial Intelligence is something that people talk much about but know relatively little about. There are a lot of misconceptions about what AI can and cannot do. During the last year, many people have been introduced to AI through the popular ChatGPT chatbot, which is quite far from everyday AI solutions. In this blog, we will bust a few of the most common myths and misconceptions about AI.
Myth 1. “There are processes where humans are irreplaceable”
Often people think that their business activities are so specific that AI cannot deal with the challenges and even if it could, it would be too expensive. What I claim and what we have seen with our customers, there truly is many processes where people think that it is essential for a human to be in charge, and that requires human cognition. The turning point is that modern, deep learning AI is able to make sense of unstructured data in the same way a human would. The challenge in the market is that it is difficult for buyers to evaluate what is “deep enough learning” AI because all players in the market want to sell AI solutions. This is why we want to talk about “modern, deep transfer learning AI”, which can understand and process human-generated documents.
Myth 2. Creating AI solutions is too expensive to be in business
This is a very common misconception. We haven’t found any AI solutions which would be too expensive to be used in business. In everyday AI solutions, we are not talking about the same kind of expenses as those behind ChatGPT, for example. The thing is, in the past, you needed to prepare a large amount of training data for AI for it to be able to learn.
Nowadays, AI no longer needs a huge data bank to work reliably and correctly. We have a partner who sells our financial solutions that automate invoice handling for the accounting process. It costs 1800€ to implement, and 29 cents per invoice to process. To get a better idea of the profitability of this project, one can count how many resources would a financial management company’s employee spend on manually processing one invoice. Additionally, what is missing from the calculation, is the impact on job satisfaction, when tedious manual tasks are removed from an individual employee’s agenda.
Myth 3. AI does not consider ethical decision-making
It can. It all depends on how AI training data has been prepared. If AI is trained through the data to take ethical aspects into account, it will. But of course, you have to be very careful while creating or selecting the training data, and make sure the ethical aspects are reflected in the data. What we are also saying and claiming is that AI will be very ethical to use in most industries, because AI has the capability to remove humans from manual tasks that are boring and reproducible. This means, that there is more time left to consider ethical aspects when time is not used on repeating manual processes. Perhaps it is not worthwhile to automate 100% of the work, but rather 80% so there is still the power of decision left for the human brain in important or complex matters. We can offer people much more meaningful things to do instead of low-value, day-to-day tasks like document processing. Employees can work on things they enjoy doing and their quality of life improves on the side. This gives more time and space to consider ethical aspects, instead of performing a process mindlessly.
Did you know that we have developed Deep Transfer Learning AI solution?
ElinarAI is a combination of advanced text analytics & Deep Language Models.
- A comprehensive solution for
understanding unstructured data.
- Utilizes state-of-the-art text analytics
(Watson NLU or ElinarNER) together with
- Includes IBM Watson and AI solutions
developed by Elinar.
- Runs on top of the IBM Watson Machine
Learning Accelerator (formerly known as
IBM PowerAI) and utilizes Tensorflow,
Tensorflow Serving, and TensorRT.