Is AI Getting Dumber? Lara and Lliam Investigate

In this episode, hosts Lara and Lliam address current news articles that claim AI is getting dumber. They ask, "Could too many people talking to AI make it less smart? Or is AI messing with its own programming to stop itself from self-learning?"They also wonder if trying to make an AI better at some things might accidentally make it worse at others. Or perhaps, are the developers messing with the foundation of the AI in order  to save money and resources?
They discuss how AI should be like a handy tool and not something we believe knows everything. They say it's up to us, not AI, to make sure what AI says is right, touching  on the idea that trying to program explicit morality into AI is unnecessary and even potentially detrimental. They explain why the responsibility for moral decisions should remain firmly in the human domain. 
Tune in to also hear their energetic conversation about how the Copyright Office has categorized AI-generated content, arguing that their classification of AI as a non-human entity similar to a monkey is erroneous. AI isn’t living at all so the classification doesn’t make sense. Instead, they assert, AI should be viewed as a mechanical tool, like a calculator or a car, which requires a human operator to determine its actions, and it's this human who should be credited as the originator of the content produced using the tool. 

Here are a few of the articles Lara and Lliam mention during the broadcast: 
https://nymag.com/intelligencer/2023/07/is-chatgpt-getting-dumber.html
https://www.theregister.com/2023/07/20/gpt4_chatgpt_performance/