One of the problems with ChatGPT is that it could easily spread false information. It wouldn’t be done with malicious intent, but it could very well be hijacked by someone with bad intentions. People who don’t fully understand the tool’s features and capabilities could also end up with false information because they failed to check their sources.
The problem with artificial intelligence is that errors are difficult to predict and even more difficult to investigate. This raises the issue of accountability and reliability. There are many fields, such as industrial engineering, where you need to be sure of the results; you need reproducibility. You also have to be able to explain why things went wrong. That’s not something you can do with AI at the moment.
In case you missed it, here is the first part.
What do you think are the biggest limitations of ChatGPT?
One of the problems with ChatGPT is that it could easily spread false information. It wouldn’t be done with malicious intent, but it could very well be hijacked by someone with bad intentions. People who don’t fully understand the tool’s features and capabilities could also end up with false information because they failed to check their sources.
Another more practical point is that the stability and reliability of the tool on a daily basis is not always at its best.
Also, fine tuning and/or input parameters (in particular the number of tokens) are aspects that deserve to be improved for internal applications.
This is a vast subject on which I could spend hours, but ChatGPT is also limited by the very nature of AIs: the black box effect. In practice, this can also limit their adoption by society.
What do you think about the future impact of AI on the workplace? Will it make our jobs easier? Is it a threat to some professions (journalism, marketing, or even development)?
Just like any other invention or new, yet quickly evolving technology, I think AI will indeed have a big impact on the way we work. Exactly like the internet back in the day! Certain jobs, such as content editor, owe their existence to the internet. Of course, there will always be people who are resistant to change in general, but I think most will adapt. That said, our ability to adapt often varies depending on our age and personality.
How will artificial intelligence change the workplace? I think it will mainly impact the way we manage our time by allowing us to accomplish daily tasks faster. All professions can be affected. Personally, I started working as a Data Scientist in 2015, my job doesn’t look at all like what I was doing at the beginning, because a significant part of the models I was developing myself are now standardized and available “out of the shelves” and many tasks are now automated by AI.
However, AI doesn’t eliminate my value as an employee, but rather changes the way I do things and organize my work. I think it makes us question the value of our work. If AI allows me to avoid unwanted tasks and focus on the ones I really want to do, then it’s a win-win as far as I’m concerned. That said, I don’t want to be naive and ignore the fact that some people will find it easier to change their work methods than others.
Can you think of any other impressive AI-based tools? Would you like to tell us about them?
The field of artificial intelligence has been getting a lot of attention. There’s something new every day, such as models, for example. I think we need to look at it from various perspectives to really get a full picture. and to pay attention to the scientific process and its temporality. Moreover, I also have a personal bias for AIs that are developed with a clear vision: what interests me is precisely what is the added value of this AI. What does it bring me as a human being, as a society? I am not someone who is easily impressed by progress for progress’s sake. What interests me is what we do with it, to what extent it helps or not.
And so from that point of view, I’m very interested in AI in the field of medicine. For example, there are AIs that will become better than the human eye at detecting cancer or AIs that are now being used for prostheses for people who have lost limbs. Another area that I find really interesting is fraud detection. For example, I had the opportunity to see research in AI using graph theory, one for the detection of suspicious vehicle “routes” (in the context of the fight against terrorism and organized crime) and another on bank fraud, it was so fascinating! In short, in all sciences, there is so much to contribute just by discovering new avenues of exploration! So many opportunities that could contribute well to our society.
Finally, could you tell us what Uzinakod has to offer when it comes to AI? Also, how is it more accessible and easier to integrate for today’s businesses? How can they benefit from AI?
I think it’s important to understand that a company like OpenAI has the hardware and financial ability to train a model like ChatGPT. Not all companies can pull this off. Also, using AI in-house can be very difficult. If it’s not done properly, it can cause problems and may not provide the added value you were hoping for.
As a company that works in AI and is used to seeing projects through to completion, Uzinakod has the expertise to identify these issues. One of our strengths is knowing the difference between fact and fiction when it comes to AI. We’re also in tune with the reality of our clients, which leads us to evaluate our contribution and the ways in which we can deliver results.
Clients often come to us wanting the best possible predictive performance, but we quickly realize that what they really want is interpretability once the model is in production. On a day-to-day basis, it’s often difficult for me to choose between these two criteria. One model is more efficient because its equation is more complicated. Unfortunately, that sometimes means that no one understands it, so it’s harder to interpret. We understand this type of problem and are able to identify the ideal model and the best type of AI to meet a company’s needs.
Want to take your business to the next level by integrating artificial intelligence into your daily operations? Get in touch with us! Sophie and our team of experts are here to help you find the right solution.
