Your donor presentation + fake AI statistics = disaster

With so much uncertainty, I wanted to find a way to add value to you, non-profit leaders, with no intent to sell you anything here. I figured I'd create these Weekly Emails I call Monday Motivational Minute because Mondays are historically the roughest days, and a minute . . . well because you're very busy people. In these, I'll add creative, quick, actionable takeaways for you to try (All in a minute read). No pressure, no need to respond, read at your leisure or unsubscribe if it doesn't add value.

---------------------

Happy Daylight Savings Monday!

Trust! That's exactly what shattered the other day when I met an amazingly brilliant lady who asked me how to stop AI from lying to her; trust that I'm supposed to be this AI expert, trust in AI overall. To be vulnerable, I went completely blank. I could not think of a single answer! Then I had to run off, like Cinderella at 11:59PM, to a consulting meeting to train a nonprofit leader on AI, and it hit me harder than Will Smith hit Chris Rock (Yeah, that hard! 😣)

What about YOU! Those who don't trust AI, with the fear it'll embarrass you by making up statistics, making up grant opportunities, or confidently lying about "facts" you share with your board.

Nope, not on my watch! I can't have you cite an impressive statistic AI gave you, during a donor presentation, only to find out later it was completely false.

Your credibility? Destroyed. Your trust in AI? Gone.

Here's the truth: AI is like that new intern that wants to impress you so much it would rather make up an answer than admit they don't know something. It's programmed to please you, even if that means gaslighting you with confident hallucinations.

(Says a lot about how much we love to reinforce our own beliefs, so much they trained AI to do it for us 😣)

So, to that brilliant lady at the coworking space and to you, here are your anti-hallucination strategies:

Daily Protection Strategies:

• Use Perplexity.ai for research (automatically cites credible sources)

• Add to your prompt: "Verify all data online and state your sources. Only use real, credible websites as resources."

• Before sharing the end results, ask AI: "Review your response as the world's best fact-checker. Identify any potentially false information and mark it UNCERTAIN; FALSE, or RESOURCE DOES NOT EXIST."

• Also, Remember: Longer chats = more hallucinations. Start fresh conversations for important work.

Actionable Task: Also Stop AI lying with one simple setting; update the brain!:

1. Step 1: Click your ChatGPT profile icon → Settings → Personalization → Custom Instructions

2. Step 2: Under "How would you like ChatGPT to respond?" add:

   "If you are not 99% confident about your answer or if you can not find any factual resource, say 'I DON'T KNOW' in all caps. Only cite real, verifiable websites. Never fabricate information."

3. This trains AI to admit uncertainty instead of confidently lying to impress you.

Caution: With all this, still verify anything you share in public. . . . with comfort I say, this is why AI can NEVER fully replace humans.

Remember: Build relationships with AI like you're co-piloting, not autopiloting. Trust speeds you up. Verification keeps you safe. The dance between the two is where innovation meets responsibility.

Regis Arzu  (he/him/his)
[email protected]
ETS AI Consulting | CEO
347.748.5078

P.S. Know another ED holding it all together with grit and grace? Forward this their way. We’re not meant to lead alone, and together, we go farther.

[Subscribe Link]