??ChatGPT my new assistant, explained to my child.

My son is 9 years old, I started to show him Chat GPT this weekend. Here is why.

ChatGPT is just amazing. Generative and conversational AI had been in the landscape for a while, but this new technology takes it to a whole new level with its advanced capabilities and high level of accuracy. It emerged very fast (it’s only 5 years from 1rst paper research to large scale beta mode) and it has the potential to revolutionize many aspects of our personal and professional lives.

However, as with any AI technology, it's important to understand its limitations, especially while the solution is still in its early days. ChatGPT can embed falsehoods in its answers, and those errors can be very difficult to predict or to identify. And as any AI system, it lacks moral compass and may generate questionable answers from an ethical standpoint.

Because of this, it's important to verify everything this technology says and use it with caution. We have to be wise to use it as an assistant, that may be very productive,?an assistant that may extend or augment us while keeping critical sense and self-reasoning.

It means that we still desperately need human expertise, and that critical thinking is more important than ever. If you have children who are likely to use ChatGPT or similar systems, have a talk and explain these systems’ deficiencies and limitations very clearly. The instinctive reaction from my son this morning when testing ChatGPT was “he is sooo cool” (“il est trop sympa!”).

First he immediately considered the tool to be a person, like a new friend. And then he wanted to test him again and again before concluding that ChatGPT was kind of his new god that would answer all his questions in the future.

I took quite some time to explain him the risks and pitfalls of such AI. I wanted him to understand that he was interacting with an AI, what an AI is and what it is not. And I know I’ll have to explain it again.

But beyond the issue of young teenagers mis-using the tool, there is a much broader and systemic risk that ChatGPT could further exacerbate polarization in society and widen the existing digital divide, as people without sufficient means may in the future increasingly rely on free questionable AI-generated content, while those with means will have access to accurate and reliable information being verified by subject-matter experts.

And whatever happens, ethics will for sure be at the heart of the debate.

Transparency of algorithm and data source will be key to prevent the tool from being injected fake news and fake data, or to guarantee that new data and new scientific knowledge is properly embedded. Transparency again will be critical to know who is driving the tool and who is deciding upon its ethical guiderails.

From a professional standpoint, lots of questions are also raised. We are starting to develop guidelines and coaching videos at Orange Business Services that will soon be broadcasted to our employees to support them in using ChatGPT in their daily work (thanks to Philippe Ensarguet who has also reviewed this post!).

The potential for better productivity is tremendous. My legal team has started to look at it and has been astonished. And all subject matter experts, whether they are from legal, HR, coms or development teams, have the potential to save huge time while asking ChatGPT to generate drafts in their fields of expertise. But the tool will not replace our experts, it will rather assist them by taking on basic tasks and allowing them to focus on higher level expertise.

We will need human intelligence more than ever because our experts will have to review generated texts with great attention and amend what needs to be corrected or complemented.

Because the model does not think, which means that you should never ask an AI to take decisions without appropriate human supervision. And while the tool progresses, ChatGPT could be to knowledge workers what automation was to the blue collars in the 1970’s, bringing new ways of working, more productivity, less work, and progressively pushing the boundaries toward higher level of expertise required into the teams.

Personally, I have started using ChatGPT as an alternative to Google when searching for specific information that does not require fresh data. I have also started to test the tool to generate new ideas and look at things from a different angle.

I have also used it to generate drafts for speeches and posts (like I did for this one ??), which saves time although multiple iterations, rework and reflection is still required. And such timesaving has drawbacks as well. We should be very careful about what we put in the prompt, as the tool is clearly bearing the risk to disseminate some of your company’s confidential information or even proprietary IP in case of development teams using ChatGPT to produce code.

It will also raise questions about the environmental impact of such solutions.

Or questions about the impacts in terms of digital sovereignty outside of the US. It raises also questions about the future business model of ChatGPT, which is very unclear at this stage. The open-source community will rush to try making ChatGPT’s underlying technologies available to all, but I doubt that the ChatGPT code itself will become open source. The model is likely not to stay free of charge on the long term.

At the end, whatever the name of the tool winning the generative AI fight will be -ChatGPT may benefit from its 1st mover advantage, but may be not-, we are clearly at the beginning of something new.