Chat GPT is becoming a favourite internet game. It has serious possibilities for learning, writing, and cheating.
The AI writer generates errors, mostly because time for the current version stopped in 2021. For example, it thinks that Boris Johnson is still prime minister – or has it been hacked by the BJ camp?
Development of artificial intelligence has been underway for decades. From primitive beginnings, it has been growing in power and in “humanness”. Contact your bank or your council and in many cases, you’ll be talking to AI by voice or online. But no one thinks these have intelligence. Mutter something unexpected to your bank’s bot like “which side should I butter my toast” and you can cut through to a real human operator. At least I think it is a real human operator.
Chat GPT, released to the public a few weeks ago, is remarkable and some commentators think it fulfils the Turing Test, which is passed when a computer’s responses cannot be distinguished from those that would have been made by a human. However, Chat GPT itself is dismissive of the idea:
“It is difficult to say whether Chat GPT, or any other language model, would pass the Turing Test.”
AI is potentially a powerful tool for politics. Could we replace phone banking with AI bots calling? Could we get AI to write campaign literature?
I asked Chat GPT: “Tell me about party politics in the UK in 750 words”. The results are impressive. It would pass as a student essay despite a couple of errors. I also asked the bot to write a poem about the Liberal Democrats. It is remarkably good if close to doggerel.
By the way, Chat GPT tells me which side you butter your toast is a matter of personal preference, including whether you butter it both sides. Does anyone do that?