How to bring artificial intelligence to your business

Artificial Intelligence (AI) is one of the most versed technological topics of recent years, a theme that has generated a multitude of debates about pros and cons, advantages and disadvantages, among other important facets, and of which course it can take is not yet known in concrete.

The fact is that in the last decade there has been an increase in investments in AI, mostly carried out by private equity or venture capital funds (75%). France (165 agreements worth 1357 million dollars), Germany (140 agreements and 520 million dollars) and Denmark (21 agreements and 330 million dollars) are the countries that invest the most in AI.

In this article, we propose to analyze the artificial intelligence and how to apply it in the companies in an efficient way.

What is Artificial Intelligence?

The first definition of AI dates back to 1956, referring to the ability of machines to think and learn in a similar way to humans (“Science and engineering of making intelligent machines”, John McCarthy).

Nowadays we have a simple definition of AI: systems that perform actions that, if carried out by humans, would be considered intelligent.

This definition is more in line with current developments than the first given definition of AI.

AI-associated tasks:

  • Sensing
  • Problem-solving
  • Language learning
  • Robotics and control

Every day we get messages that AI is going to change everything, that it is prepared to change everything, putting many jobs at risk. Many people are afraid of becoming obsolete.

Artificial Intelligence in Portugal

National companies are already aware of the advantages and importance of artificial intelligence. The “Artificial Intelligence in Europe” study of the Ernst & Young consulting firm concluded in 2018 that:

  • 91% of the leaders of organizations in Portugal expect the IA to benefit the business through the optimization of operations.
  • 77% believe that this technology will be key to engaging customers.
  • 73% believe that AI will release workers from repetitive tasks to engage in value-added activities.
  • 55% believe that it will guarantee improvements in the transformation of products and services.
  • 45% of Portuguese companies have not yet started pilot projects with AI to improve their processes and tasks.

How to implement Artificial Intelligence in companies

Much is said about technology, but no one talks about how to implement this same technology. How can artificial intelligence work in companies?

Here are some tips on how AI can materialize in companies. Nowadays everyone talks about AI, and many people want to apply AI in their activities, but we are forgetting the classic, which is very important.

In summary, there are three great steps to take by companies that decide to move forward on having some product in the AI area:

  • Don’t focus on the technology, focus on the functionality: we have to develop applications that are functional and that solve the end-user problem.

When it comes to addressing this issue, the first question a company has to ask itself is whether if it is functional to apply the AI for this. This functionality is supported by algorithms and data, but the algorithms are only one instrument.

  • We must take into account that AI is many things, not just deep learning, not just machine learning. We have several areas that support AI.

It is important to know which of these technologies that support AI will be used to develop the work.

 

  • To try to have products in this area it is also important to think of 3 issues:
  1. Task – is your task genuinely data-driven?
  2. Data – will the data we have suffice to support the approach that will be taken?
  3. Scale – do you need the scale automation provides? Is there capacity to expand?

Universities also have an important role to play in AI. New technologies are important, they are powerful, but universities should also be concerned with teaching students how to design and build the AI of the future. At present, the major focus is on technologies and not how to think about these activities for the future.

Now we have a deep neural network but in the future, there may be another technology.

Finally, AI requires integration, that is, we must have learning, we have to connect business and technology, ideas and their execution, linking the thought and the engineering (which is the way to connect it all).

It is important that when we build an AI model we are sure to explain, that we have some idea of how we came to that conclusion. Make sure that the intelligent systems brought to your workplace are able to communicate not only the answer but also the reasoning and the data that supported it.

Metrics to consider

Examples of metrics to take into account when identifying and solving a problem using artificial intelligence:

  1. Let business do the driving.
  2. Think of tasks not jobs.
  3. Understand the problem and the goals.
  4. Evaluate the data from a human perspective (and not only what the algorithms say).
  5. Make sure the data are actually available and have quality.
  6. Understand what is being learned and how it will be used.
  7. Ask whether the target is learnable given the data.
  8. Consider how the solution will be integrated into the workflow.
  9. Start with simple design against near term ROI.
  10. Consider the impact of regulatory requirements.

About the author:

Miguel Guevara | Development coordinator @CCG, D.I.A. CVIG

Miguel Guevara has a degree in Mathematics and a Ph.D. in Technical Sciences. He is Senior Researcher and the Development Coordinator of Applied Research Domain CVIG. He has experience as professor and researcher of more than 28 years in the area of Computer Science and has published a total of 85 papers/articles of a scientific nature.