There are many articles, videos and keynotes about AI. Often wondering what it is, if we can trust it, if it is taking our jobs, if we should be afraid of it… I decided to join the discussion and give my perspective on it.
What is it? A very knowledgeable, experienced and super fast partner. Always willing to help through sparring and giving inputs or suggestions, allowing us to be more efficient by doing manual work way faster than we do, or even allowing us to perform tasks we would never be able to perform given the volume of it.
Can we trust it? well, no, we should not. As we would not trust anybody blindly and believe anything they come up with without verifying what is being said. Even if people seem experienced and knowledgeable, we would keep our critical sense and check the truth in what they say before making our opinion and deciding to trust the information they gave. Same with AI, whether we use it for sparring, getting condensed information or even for vibe coding, we should always be the one in control. Even more so, as we know that the way AI is built, it will eventually lie or hallucinate. Trust needs to be built through strict processes around it. We give a command, build AI models, we get some outputs (in no time, that’s great) and we get to quality control.
Should we be afraid of it? Not of « it » as such, but we should be afraid of people not using it properly. If we start trusting it blindly, not questioning or verifying what it does, not building strong processes and governances around it, then that is scary. In that case, we will have no control and have no clue of what is going on. We would never advise a student to give back to their teacher a paper that someone else wrote for them without checking it and knowing what it is all about. In the same way, we should not let AI do all the job without us knowing what it contains, what inputs it is using and how.
Is it taking our jobs? Not really. It is contributing to making our job evolve and forces us to have another approach to it. It won’t leave us jobless, but we will spend time on different tasks, assessing more than producing. Quality controlling information more than gathering them ourselves. We will spend more time on what makes us special and increase the value we create. We will spend more time on making sure AI gets the right inputs to give the right outputs (governances, structure, data quality). AI accelerates our work and increases the quality of some of our outputs but it does not replace people.
Implemented well, AI is a powerful tool, able to support every one of us being more performant and efficient.
What does implemented well means then?
Clear roles and processes: understand the role of AI and the role of the people.
It means that a process involving AI requires, as any other processes, to have clear roles… including our new friends role. The role of AI is to gather, compile and condense information or execute a manual process faster than we would. The people’s role is to give AI the frame to do it in. People tell AI where to look and what to do, we quality-control and challenge the output until we are confident it can be trusted. As an end user, people need to make sure to be clear and concise in what they ask, we need to be critical and verify the source of the information, we need to ask AI to justify their answers or their code and be on top of what AI comes up with. AI is great, learns fast and help people save a considerable amount of time. Coming back to what it is, it is a very strong partner for people, but what it is not is an independent worker able to perform a task and give the output to let people free from thinking. We have the main role in the process of working with it.
Don’t go too fast
One of the great aspects of AI is that it is not a static product, it is learning as people work with it. When launching AI solutions, we need to keep that in mind. Start slow, try it out. Like in any new partnership, we have to build the trust until we can rely more and more on it (within our governances and framework). We should always be aware that it can be wrong or hallucinate and we need to take the time to create the space in which it can be trusted. By taking that time, we are building our own mechanism to catch when it is wrong and enabling us to be confident in working with it.
The people using it, decide how and when to use it and are in control of the output we « co-create » with it.
As for any technology, implementing AI should be focused on the people and how they should (and not) use it.
- Be curious, but stay critical.
- Appreciate the help, but stay on top.
- Build the frame.
- Build the partnership.
Only then, AI is a strong partner.
Claire Lemettre