AI hype: Do you really need AI to solve all your problems?

In recent years, AI has maintained its position as one of the most promising and widely discussed technologies. Interestingly, it attracts the attention not only of technical experts but also of people far removed from the world of technology. Why is this happening? What is driving the hype around AI? To discuss these and many other questions, Maxim Golikov, the host of the Innovantage podcast and the CBDO at Sigli, invited AI experts to his studio. The guests of this episode were William De Prêtre, Head of AI at AllKind Group, and Artem Pochechuev, Head of Data and AI at Sigli.

Both of them have been working with artificial intelligence for many years. During their careers, they have observed different stages of AI development. For this episode, they agreed to share their vision of what is happening with this technology now and what we can expect to see in the years to come. They also explained the key challenges that organizations may face when integrating AI into their solutions. These insights will be of great help to everyone who is considering the implementation of AI in their companies now or in the future.

Education as the major step toward AI introduction

As both Artem and William have incredibly rich experience in working with AI projects at their companies, Maxim asked them about the most important preconditions for successful AI implementation. Their answers may seem surprising to a huge part of the podcast’s audience. Both experts mentioned that the first thing that should be done before bringing AI to people is educating them on what AI actually is. If you just ask random people about their understanding of artificial intelligence, they will say that it is ChatGPT. In reality, AI and its use cases go much beyond this.

The problem is that today a lot of people who want to use AI have very limited knowledge of this technology. As a result, they can’t find the best application for it. However, using AI just because that’s AI is the wrong way.

AI itself has become very efficient. But it is not necessary to apply it everywhere. A lot of solutions can work without it. According to William, if you can solve something with just your high school statistics course, then solve it with this knowledge and not with AI. This will let you use your AI resources for something that really requires AI.

The term “AI” has become a powerful marketing tool. You can perfectly sell something by just saying that it has AI even if it doesn’t use this technology at all.

As Artem noted, the first thing when it comes to decision-making regarding the implementation of something new should be awareness. To adopt something, to decide that you need something, to start planning something, you need to be aware of that. That’s why this education should be company-wide. Not only potential users but also decision-makers should be educated on tech-related questions.

The second thing that you should focus on is the process of AI implementation. To implement this technology, you can’t avoid having tech-savvy people on board. These people should be aware of AI and be ready to go deeper and deeper into AI topics. A lot of businesses prefer to have a reliable technology partner. Or they have a choice to grow their own engineers who will be able to cope with all the required AI-related tasks. Moreover, there should be specialists who will help the company define the right purposes and priorities for their AI projects.

How to introduce a new tech solution

At the same time when you bring something new to managers and want them to let you implement some new solution, you should be ready to show them the full potential of this innovation. It is vital to explain everything in simple terms in order to let everyone fully understand your ideas. Managers do not need to know technical details. But they need to know what value they can get with the introduction of some new technologies.

It doesn’t matter who will bring a new idea to the table: tech experts, business people, or external partners. What does matter is how people adopt it. How do they understand it? How quickly do they apply it to real work? How do they avoid potential risks? It doesn’t matter where exactly the implementation process starts. It is much more important how you continue with that.

William also mentioned that the success of solutions often depends on the contributions of different teams. His company builds innovative products for students with different needs, like the Web2Speech extension that can read text content aloud. That’s why the success of such projects is preconditioned by the interplay between input coming from engineers, input coming from people from the education sphere, and input coming from management. There is constant interchange, which is required to have success in the AI market.

Main challenge

Speaking about edtech solutions, William highlighted one very important aspect that many people can forget. Such solutions deal with children’s data. That’s why privacy laws, GDPR, the AI act, and other related regulations become very important.

Intuitively, you may know that anonymizing your data is crucial. But practically, this will greatly complicate a lot of things for you.

However, you can’t avoid taking care of data protection. It is really necessary because your solution will work with tons and tons of very sensitive data. And of course, you can’t let it leak because this situation will kill your reputation.

The more widespread AI becomes, the more attention companies need to pay to data security and privacy protection.

Unfortunately, that is something that engineers tend to overlook because they are focused on making AI perform in the right way and may forget about the value of some data for people.

How to get ready for working with data

According to Artem, quite often, people underestimate the significance of data in AI in general. However, it plays the most important role. If there is no data, there is no AI. Without it, you can’t train your AI/ML models that grow into a large language model (LLM). You can’t train anything if you have no data. That’s why data comes first.

One of the most crucial steps that are required for AI adoption is shifting to the data-centric direction. Unfortunately, that’s exactly what companies often miss.

Of course, a lot of people have heard about AI but they perceive it as some kind of a jack-in-the-box that can just jump out and do everything you need. But it doesn’t work this way. AI should be trained with data before it can do anything for you.

In this context, William mentioned one of his company’s projects known under its code name Bulbasaur. It is an AI tutor that can assist teachers. It can be fed with course materials. And namely, these materials and their quality will show how good your tool is. If the solution doesn’t have enough data, it will not be able to answer your question. But this situation doesn’t characterize your solution itself. It just shows that you haven’t provided it with relevant data.

Without sufficient data, it simply won’t work.

This principle is applied to any AI-related task. It doesn’t matter whether we are talking about predictions or clusterization. All such tasks will be performed on data.

Even if you want AI to reformat your presentation, you need to feed it with your thesis, abstracts, and other presentations first.

How to define your AI needs correctly

Artem explained that he usually splits the AI needs into two categories: an internal track and an external track. An internal track is all about tools that can help your employees perform their usual duties more efficiently and bring more benefits to the company. Another thing is projects that you as a company sell to your customers. Here, it’s important to understand whether you can improve your projects with AI tools.

Being at the crossroads as a decision maker you need to choose which way to go. It’s vital to clearly detect the pains of potential users. This will give you an understanding of the exact tasks that your solution should perform. At the same time, you also need to talk to engineers to gather their opinions on how such tasks can be solved with tech solutions.

Nevertheless, the introduction of innovations does not always go smoothly. William said that you can also face resistance to change and it’s not just because you are offering AI solutions. In his practice, there were similar cases with cloud services. When his company started moving solutions to the cloud, a lot of customers were quite confused by such a decision. Nevertheless, now people complain quite a lot about their non-cloud solutions.

Given this, it’s possible to assume that at some point somebody will be not satisfied with tools that won’t have AI.

You should also be ready for situations when it is not feasible to continue a project that seemed to be a good one at the beginning. It may happen because there are not enough resources for it or because it is not fully supported by your company. William advised not to throw it away but to put it aside. It may be still viable sometime later and you will be able to return to it.

Practical tips for AI implementation

At the end of their discussion, Maxim asked the experts to share their recommendations with those who want to start their journey with AI.

“Surround yourself with good people. Educate everybody. Find good partners,” William said. He recommended exploring all available options. “The path to heaven is clear. Go ahead and build your ladder. So even if you’re not in heaven yet, at least you can hear the angels singing,” he added.

According to Artem, the best way is to grow professionals inside a company and get their expertise. He explained that today many people are ready to train you to work with AI. But in reality they just want to get easy money. That’s not what a successful education is. You need to have a decent person who is able to go deep and share the knowledge. This is the most effective way to educate people all around you and people in the company.

William also highlighted the importance of industry conferences and organizations that support tech companies. Sometimes they can provide funding or help you get into contact with the right people.

AI future: What is it?

It’s interesting to see how people’s opinions about AI are changing over time. Initially, teachers voiced a lot of concerns about children using ChatGPT for their homework. Now some teachers in Belgium explain to high school students how various types of AI work and help them build small AI projects using off-the-shelf components. All this indicates that quite soon we will have a new generation for whom AI will be just part of their everyday life.

Of course, it’s very hard to predict the future, especially in something that is moving so fast as AI. Nevertheless, it’s possible to make some general assumptions based on what is happening now.

For example, according to William, there will be far more autonomous systems and self-driving cars. But he doesn’t think that they will come from Tesla as there are other car manufacturers that are already far more advanced in their autonomous technologies. Apart from this, there will be more autonomous drones used for military purposes, as well as AI personal assistance agent systems, in which small dedicated agents will work together to solve bigger problems.

Nevertheless, William hopes that we won’t see more AI-generated images in the future. According to him, an AI-generated Hollywood blockbuster won’t be the best idea. He said that we should assign our boring tasks to AI, while more creative, fun work should be still performed by people.

Artem added that we should perceive generative AI as a tool, not more or less.

As for predictions, he also said that it is quite useless to make them. Right now, there are a lot of talks about AI hallucinations but 3 years ago we didn’t even know what it could be.

That’s why when somebody is trying to invent any framework protecting us from the vicious AI of the future, it is mainly just a waste of resources. The future may turn out to be different from what we expect now.

Wrapping up

Artificial intelligence is a highly potential and powerful technology that, with the right approach, can help us solve many tasks of different types.

However, as the experts advised, we shouldn’t use a microscope to hammer nails.

Today there are plenty of things people are trying to solve with AI. But in reality, such things do not need any sophisticated approaches and can be solved much more easily.

One of the core things required for successful AI adoption and implementation is comprehensive education of people on the basic questions related to this technology. It’s vital to know what AI is and how it can be used to bring benefits to your organization.

The Innovantage podcast has a similar role. It helps to increase the awareness of the audience on various business and tech topics with a focus on AI and its capabilities. If you want to learn more, do not miss our next episodes!