What has astounded me so much about OpenAI’s ChatGPT is that it really seems to understand the meaning of what you say to it. But is this real or is it just a parlor trick?
Past AI chatbots have relied upon trickery. They search input text from the user for particular words and use if-then conditionals to provide canned responses. ChatGPT is clearly doing more than that, but that doesn’t necessarily mean it truly comprehends language.
Time to experiment.
- Can ChatGPT identify grammer?
- Make inferences?
- Identify nonsequitors?
- Identify nonsense?
Continued below the fold…