Back to Main
This Week in AI — ⚖️ When does AI use become unethical?
AI

This Week in AI — ⚖️ When does AI use become unethical?

February 20, 2024 | By: Dave Isaac

Headlines You Should Know

 

New AI Tools, New Ethics Questions

There’s a quote that goes back to the French Revolution, although it’s often attributed to prominent world leaders and, uh, ::checks notes:: Spider-Man. With great power comes great responsibility.

The notion holds true for artificial intelligence (AI), which has been abused by hackers for cyberattacks and bad actors looking to influence elections. AI’s evolution has made it easier to create lifelike videos and replicate a voice using text-to-speech tools, substantially reducing the time it would take to create a finished product manually. It becomes even more powerful when you pair OpenAI’s Sora model with AI-generated sound from ElevenLabs. Those innovations raise new questions and blur the line of ethical ways to use AI.

For instance, what side does using AI to regenerate the voice of gun victims reside on? Or The Lincoln Project recreating Donald Trump’s late father to berate him in an ad?

These are the kinds of questions companies will grapple with as they use new technologies to craft communications. There’s no one-size-fits-all answer that will work for every organization, but one best practice is to disclose the use of AI. That way you can at least build trust through transparency with clients, partners, and other stakeholders as you consider and document where your organization will draw the ethical lines.

Elsewhere …

Tips and Tricks

🧠 How to control ChatGPT’s memory

What’s happening: Last week, OpenAI announced a new feature (not widely available yet) that lets ChatGPT remember commands, preferences, and details across all your chats. It sounds a lot like custom instructions, a feature that debuted last July, though more automated.

Why it’s important: The ability to give ChatGPT a base set of instructions or understanding could save a lot of time repeating information and instructions. It gives ChatGPT more context, so it has a better starting point and makes it more likely that the output is what you’re looking for.

Try this: For communications professionals, it’s all about the words we choose — and typically, we wouldn’t choose the same ones ChatGPT does. If you don’t yet have access to the memory function, you can try custom instructions to accomplish the same thing. There’s potential to use these functions as a way to capture a specific voice, tone, and style, but it may be best to start small and just avoid signs of typical AI writing.

For instance, my custom instructions read, DO NOT UNDER ANY CIRCUMSTANCES use generic openings like “In an ever-changing world” or “the evolving landscape of…” or use writing crutches like “it’s not only XXX; it’s YYY” or metaphors like “beacon, transformation, tapestry, etc.”

All writing should be professional and layman-friendly. Use straightforward language and varied sentence structure.

Quote of the Week

“If you put me to a wall and forced me to put probabilities on things, I have a sense that our current remaining timeline looks more like five years than 50 years. Could be two years, could be 10.”

— Eliezer Yudkowsky, AI researcher, to The Guardian on humanity’s future in an AI age