Nine Million Penguins?

According to the New York Times, testing has found newer AI models hallucinate at rates as high as 79% (Megan Liscomb, 15 Hilarious Google AI Fails That Made Us Laugh, 2025).

Did you know there’s a proper etiquette for giving your AI assistant feedback when its answers turn out to be obviously incorrect?

The internet lists numerous examples of AI systems “hallucinating” or producing an answer that is completely wrong because the response to a question has no basis in reality and isn’t anchored to real data, sources, or logic.

From declaring that 9.3 million penguins live in Rome (Oops, My AI Did It Again | LinkedIn), to making up legal cases and citing them in briefs (Oops, My AI Did It Again | LinkedIn), to fabricating a civilian hero during a shooting (AI Incident Tracker), to claiming that adding gasoline to spaghetti while cooking makes it spicier (AI Incident Tracker), AI mistakes abound.

My own virtual assistant gave me the wrong information twice in a row this week when I asked it a simple question about the schedule for a baseball game, and I had to tell it where to look for the correct information. Of course, since the AI is an automated system and not sentient, it couldn’t be either embarrassed or annoyed when I pointed out its mistake. But it was overly apologetic, and that made me wonder if there were etiquette rules for working with AI. Here are some suggestions I found online.

Etiquette rules for how to treat your AI assistant (from AI Etiquette Guidelines for the Workplace — Emily Post):

1. Understand the capabilities and limitations of your AI system. “Gaining an understanding of the strengths of the system, such as data analysis, scheduling, or creative assistance, can help establish realistic expectations while also mitigating frustration associated with any limitations.” (Or in other words, be prepared for it to occasionally be wrong.)

2. Keep a professional tone when working with your AI. “While it may be tempting to engage with AI in an informal manner, it is essential to uphold a professional tone in all workplace communications. Utilizing appropriate language and avoiding slang or overly casual expressions sets a standard for respectful interactions.” (So you should probably avoid swearing at it as well.)

3. Acknowledge contributions. “When utilizing AI-generated outputs, it is advisable to credit the tool when appropriate.” (So that other humans will know you used AI, not because it will hurt its feelings.)

4. Provide Feedback. “AI systems improve over time with user feedback. If you notice inaccuracies or areas of improvement, politely provide constructive feedback, just as you would offer suggestions to a team member.” (Really? Are you sure we’re not worried about hurting its feelings?)

Etiquette rules for using AI at work (from Can AI etiquette Boost Trust, Not Risk?):

1. Verify facts. “Find and read primary sources, and link to them. This reduces misinformation and supports due diligence.” (Again, prepare for it to be wrong.)

2. Add your own perspective. “Do not copy-paste AI outputs unchanged; summarize and adapt with human judgment.” (Maybe to prove you are needed, and that the AI cannot take over your job?)

3. Ask for consent. “If you plan to forward AI text about a person, get explicit approval first.” (Because you never know where it will end up on the internet or who will have access to it.)

Etiquette rules for using AI in daily life (from AI Etiquette: Do’s and Don’ts - Diane Gottsman ):

1. Respect privacy and confidentiality. “Steer clear of sharing sensitive, personal, or proprietary information in AI tools.” (Because you never know where that will end up, either.)

2. Don’t bait and switch. “Using images that misrepresent you or your life is a red flag. While it’s fine to use AI to edit for grammar … allowing AI to create a completely different version of you will definitely backfire.” (Pretty much guaranteed.)

3. Don’t copy and paste. “Serious conversations should take place in person, such as a heartfelt apology, a possible break-up, or a sincere thank you. A copy-and-paste version is cold and insincere.” (No kidding? At least this time the concern is for human feelings.)

The end result of all this advice seems to be that you should treat your virtual assistant as a valued, somewhat indiscreet co-worker who can be—and often is—wrong. So the next time your AI assistant instructs you to add glue to your pizza to help the cheese stick better, or to eat a small rock every day for nutrition, remember to question its sources. Politely.

***

What odd answers have you gotten from your AI? Do you double check its answers with reliable sources? Send me a note on my contact page, Contact - Lucinda Gerlitz, and let me know.

QUESTION:

If you are dining at a fancy restaurant and you drop your napkin, should you:

A. Pick it up discreetly and continue using it

B. Pick it up, hand it directly to your server, and request a clean napkin

C. Pick it up, place it above your plate on the table, and request a clean napkin

D. Leave it on the floor and request another napkin

ANSWER:

D. No matter what you drop, you should never bring something that has touched the floor back to the table. Plus, crawling under the table or fishing around near other diners’ feet is far more disruptive than a quiet request for a fresh napkin.

I’ve been busy this month preparing marketing materials for when Manners or Mayhem is released. I’ve collaborated with a designer on business cards and a bookmark that I can hand out, and I’m very pleased with what she has come up with. I also took classes on preparing a media kit and a press release, and I’ve been gathering everything I need for those.

A media kit is a packet of materials that gives journalists, bloggers, and podcasters everything they need to know to talk about you. It usually includes:

• Author bio (short + long versions)

• Professional headshots

• Book information (cover, description, ISBNs, formats, release date)

• Key talking points or interview topics

• Press releases or announcement blurbs

• Notable reviews, endorsements, or awards

• Contact information (publicist, agent, or author email)

• Links (website, Substack, social media, purchase links)

I plan to be a guest on a few blogs and possibly a podcast or two when the book comes out, so I want to be as ready as possible.

Work on book 3 in my series, with the temporary title of The Doll’s House, is also progressing, and I reached the halfway point this week. One of my suspects surprised me by surviving a murder attempt, so I’m having to adjust my outline a bit. Sometimes characters just won’t do what you expect!

For personal news, Manchester’s big music festival, Bonnaroo, has been going on this week, and I’ve been treated to the sound of a bass thumping until four o’clock in the morning. We can hear the music even though the site is three miles away, and I feel really sorry for the people who live closer to the farm where the musicians perform. My husband and I pretty much hibernate for the week of Bonnaroo to avoid the traffic, and so I’ve mostly been at home writing and entertaining the cats.

The cats refused to pose for my monthly picture, so here is a throwback shot of the Chihuly exhibition at the Denver Botanic Gardens.

Hope you’re all having a wonderful June!