License - Blog Posts

Studying

Today I decided to study for my driving license.

Turns out that it’s not different from studying for school and after 2 question I suddenly had a clean room, showered and even changed my bedsheets. What a surprise, right?

Chat GPT in language learning (part 2)

My previous post relied heavily on my opinion drawing on my long experiences as a teacher, but what of outside sources, what other issues do people see with using ChatGPT for your English language communication needs?

This article highlights the risks of sharing your company's sensitive data since the service relies on machine learning. That means that the extremely powerful AI system is learning from everything we tell it. Many are questioning whether the information people have asked ChatGPT to write their reports about is suddenly available to anyone who enters the right question in the site. The site claims that one research into this probability showed that there is indeed a likelihood of it happening.

Long running tech news site ZD Net highlighted 6 potential ways that the AI service can be used. Aside from the obvious composition of phishing e-mails or clickbait articles, they also suggested that people could use the service to more efficiently gather data to help support them in committing a crime, such as finding what IT systems a bank uses. While this seems unrelated to your language needs, imagine if you are feeding company information into the interface to help explain your new technology to internal stakeholders. That is now inside Chat GPT's data resources.

One LinkedIn user asked ChatGPT the pros and cons of using it for blogging, which I also feel can be applied to using if for your language needs. ChatGPT questioned its ability to write creatively, which to me suggests that readers may be able to detect when they are answering correspondence sent by the AI engine. A lack of depth, a lack of context, and a sub-standard quality of writing may also damage the reputation of a company.

While IBM may have a bias against ChatGPT, since they would prefer to promote their own WatsonX, they do have a couple of other compelling points about the usage of the trendy AI. Intellectual property issues may come up due to ambiguity or ownership of the content rights as well as what the answers they generate are based on. Similarly the ownership of different open source libraries could be contentious and lead to legal issues.

All these are different angles from the reliability I talked about in my previous post and should also factor into a company's weighing of the decision whether to use it for your language needs. To me, there are a lot of potential issues that go on top of the questionable reliability of Chat GPT in this method of usage, but I would be at fault if I did not explore some of the pros associated with this topic. Therefore that will be next in this series of posts.