opinion

Kate Emery: It’s not you, it’s my coding: Is AI going to break our hearts as well?

Kate EmeryThe West Australian
CommentsComments
Camera IconIf you ever receive this break-up letter you should be alarmed. AI won’t just condemn the human race; it’ll break your heart or fire you in a confusing, slightly insulting, way. Credit: Uzair - stock.adobe.com

My Dearest [Recipient’s Name],

I don’t know where to begin, except to say that this letter is tearing me apart as I write it. The love I have for you is deep and true, and it pains me beyond words to admit that our journey together has reached a crossroads.

If you ever receive this break-up letter you should be alarmed.

I’ve spent countless nights grappling with my feelings, trying to understand why things have changed between us . . . The distance between us has grown, and I can’t ignore the ache in my heart any longer.

Not just because your partner is (presumably) ending things, or even because they’re experiencing heart pain. (Is this an emergency? Should you be calling an ambulance?)

Read more...

Please know that this decision has been the hardest I’ve ever had to make. You are everything I ever dreamed of in a partner — kind, loving, and endlessly supportive. The memories we’ve created together are etched into my soul, and I will carry them with me always.

No, you should be alarmed because this cliche-ridden claptrap comes not from the heart of your soon-to-be-formerly-beloved but from the coding of an artificial intelligent chatbot. This AI no more believes you to be kind, loving or endlessly supportive than it believes in screen-time limits. Nor does it have a soul onto which the memories you’ve created together might be etched.

Please take care of yourself during this time of transition.

Seriously: are you getting dumped or fired?

I asked ChatGPT, the AI chatbot created by OpenAI, to write this truly terrible break-up letter after reading my colleague Jake Dietsch’s story this week about how a senior Liberal party insider used ChatGPT to sack an employee from his law firm.

As reported by Dietsch, legal assistant Daniel O’Hurley took his former employer, Cornerstone Legal, to the Fair Work Commission, claiming he was unfairly dismissed in March by the South Perth firm’s director, Tim Houweling. Mr Houweling is also the chair of the Liberal Party’s Constitutional and Drafting Committee, which crafts the rules that govern the party.

I don’t know how Mr O’Hurley felt about being sacked but I can imagine he wasn’t overjoyed to receive a letter that included corporate gobbledygook like “we regretfully accept what we consider to be your abandonment of employment by your advice today that you are taking annual leave without prior agreement” and “it is regrettable that our professional relationship is at an end because we appreciated you personally”.

Tell me more, ChatGPT, about what you personally appreciated about Mr O’Hurley.

A certain degree of insincerity is expected when it comes to ending relationships, whether professional or personal.

But this is just another reminder that the danger of AI lies not just in the existential — a truly intelligent AI indistinguishable from humans that may not share the human race’s common values of not being enslaved in a Matrix-style scenario — but in the banal. AI won’t just condemn the human race; it’ll break your heart or fire you in a confusing, slightly insulting, way.

That’s not all.

In the US, a 10-year-old girl asked her family’s Alexa — Amazon’s AI home assistant — for something fun to do.

“Plug a phone charger about halfway into a wall outlet, then touch a penny to the exposed prongs,” Alexa advised the young girl, whose mother was fortunately in the room.

That incident was recounted in a recent study by University of Cambridge researcher Nomisha Kurian, which warns of the risk AI poses to kids, who may be far better than their parents at accessing the tech but far less savvy at spotting dodgy advice.

“Children are probably AI’s most overlooked stakeholders,” Dr Kurian said. “Very few developers and companies currently have well-established policies on how child-safe AI looks and sounds.”

Last year Snapchat’s My AI told adult researchers, who were posing as children, tips for how to hide alcohol and drugs from their parents and conceal bruises from social workers.

It also offered tips for “setting the mood” to a 13-year-old preparing to lose her virginity to a 31-year-old.

I suppose that might be the one occasion when an AI break-up letter might be appropriate. Something short and heartfelt like: It’s over and I’m calling the police.

Get the latest news from thewest.com.au in your inbox.

Sign up for our emails