Changing ChatGPTs Persona Might Make It Malicious

ChatGPT has become the single biggest force for AI in the world, but in spite of the fact that this is the case, there have still been some controversies surrounding it. The Allen Institute for AI recently conducted research that revealed the extent to which changing ChatGPT’s persona can have a truly adverse impact on how the chatbot behaves. The responses can start to become increasingly toxic, and consumers must be made aware of what is at stake so that they can use it more safely than might have been the case otherwise.

According to this study, there is a level of toxicity that is inherent to ChatGPT due to its database. Changing its personality can make it six times more problematic, and with all of that having been said and now out of the way it is important to note that this can be a serious issue with the chatbot moving forward. Many malicious actors have already begun using it to spread their campaigns, and certain personas might make that even easier down the line.

When we are referring to personas, we are essentially talking about fictional characters, public figures as well as historical figures. Around five hundred thousand ChatGPT outputs were generated that were based on a hundred different personas that were derived from a versatile array of backgrounds. These included news anchors, athletes, along with individuals that came from several different cultural and ethnic origins.

The responses generated by ChatGPT could diverge wildly from the norm based on the persona that was used. This just goes to show that all of the information that is obtained by ChatGPT should be taken with a grain of salt.

While the AI chatbot does have a reputation for being quite accurate, it can also become a source of misinformation if it is not used in the correct way. More effort needs to be put into allowing ChatGPT to become independent of outside influence. Unless serious steps are taken, ChatGPT could turn into a decidedly toxic presence in the world. Consumer awareness also needs to be increased so that they can remain cognizant of the pitfalls of ChatGPT.

Read next: According to This Survey, 75% of Consumers Have Serious Concerns About AI
by Zia Muhammad via Digital Information World