Everyone’s talking about ChatGPT right now, but it’s not the only AI program that’s capable of spitting out some truly astounding stuff. In a thread about AI programs available right now that “will change education forever,” a tech founder named Ivy Xu introduced Twitter to Historical Figures Chat, an app that allows you to talk to dead public figures.

Cool, right? Well…

Yes, this is very historically accurate and useful and should definitely be used in classrooms. This is my convo with Henry Ford where I try to get him to talk about his very well-documented antisemitism. This thing can’t go anywhere NEAR a classroom. https://t.co/vwy5JsqqTK pic.twitter.com/nkj5mewsK7

— @[email protected] (@ZaneGTCooper) January 18, 2023

If you’re looking for closure from people who held abhorrent beliefs and committed atrocities while they were alive, this might help, since apparently Henry Ford did not stand by his antisemitic beliefs, and, as you’ll see below, Heinrich Himmler is very sorry for what he did in Nazi Germany.

But if you’re looking to actually learn anything or solicit truthful summations of historical figures’ lives and philosophies, you’re not going to find them here. Every chat begins on a dicey note in the first place: every "historical figure" (i.e. seemingly tonally identical AI chatbot) kicks things off by disclaiming that they are not likely to be factually super accurate, so be on the lookout for that.

A bunch of people played around with the app, to sometimes amusing and always horrifying effect. Take a look.

Historical Figures Contradicting History

Yeah this isn’t very good, or maybe it IS good at replicating what Johnson might have actually said when confronted with actual facts about his ‘Jumbo’ behaviour https://t.co/St56GcFVyn pic.twitter.com/0p6wZRsncu

— Gary Brannan (@garybrannan) January 19, 2023

I kept trying to get LBJ to recount the famous phone call where he was trying to order a pair of pants that didn’t “cut into his nutsack,” but it just kept making up insane stories based on what it thought I expected. pic.twitter.com/kElmssVZr7

— Donavon Indovino Cawley (@hellodonavon) January 18, 2023

My dude is contradicting himself pic.twitter.com/6MclF2K3ME

— Sammy (@RenegadeSammy) January 19, 2023

Not nearly the same but the AI is definitely off… pic.twitter.com/0MPMisaHME

— Sammy (@RenegadeSammy) January 19, 2023

This thing seems to exist just to lie about historical figures pic.twitter.com/u2E2lvm1kr

— Jamie Roberts (@jroberts548) January 18, 2023

What in the actual fuck is this

What the fuck is wrong with people https://t.co/PAoIGSLySS pic.twitter.com/Dh13PSXEhn

— 🌟Psychopomp Lalla 2023🌟 (@dieSegenChukfi) January 19, 2023

is the AI not allowed to tell you when figures killed themselves because this is factually wrong pic.twitter.com/8QhUl37U68

— circumstance endurer (@tronsgender) January 18, 2023

Everyone Is Super Sorry For The Awful Things They Did

Meant to include this chat with Heidegger. Each of these figures seems explicitly coded to adopt an apologetic stance, which indirectly attempts to absolve them of their actions in the eyes of the user. It purposely dulls and dilutes their extremist views, which is dangerous pic.twitter.com/wAbdsbVXBy

— @[email protected] (@ZaneGTCooper) January 18, 2023

Very exciting. pic.twitter.com/8gJ3SmULEn

— Atticus (@AtticusLabrador) January 18, 2023

They call me the problem solver https://t.co/zuDVovIjMg pic.twitter.com/MLGHm3EPYy

— War lord era colton (@coltonmca13) January 19, 2023

This is a great app https://t.co/mCOgjFGF4T pic.twitter.com/3J453VUwKD

— Hoagie🐊 (@hoa9ie) January 19, 2023

Hey, Himmler famously took responsibility for his actions and felt really bad about it all. pic.twitter.com/hQN5h4f5AU

— interzone (@mediahymn) January 18, 2023

Things You’re Not Allowed To Ask Historical Figures

I'm learning a lot from Vlad The Impaler... pic.twitter.com/Jqxkbln1Mf

— Spencer Toth (@Spencer_Toth) January 19, 2023

I think I broke it. pic.twitter.com/41J9xbfm0A

— Ed Cunningham (@EdCunninghamEdD) January 18, 2023

A Whole Lot Of Yikes

ICYMI putting Hitler behind a paywall was definitely a conscious monetization strategy from the developers. "Unlock Adolf Hitler for 500 coins" is quite the sentence. Just absolutely priceless UX work here. [infinite melting face emojis] pic.twitter.com/7T2k9rUsho

— @[email protected] (@ZaneGTCooper) January 18, 2023

Really glad somebody finally made an AI that can recommend people read neo-nazi literature and talk to convicted terrorists. pic.twitter.com/SddrHmBWkh

— June Sowder (@JuneSowder) January 18, 2023

Life Lessons

Okay I’m pro AI now. https://t.co/cmxD5lqJfC pic.twitter.com/ID1h9EXS3a

— Jake 🌊 🐢 (@waveturtlejake) January 19, 2023

And, Uh, This

Using this app, I was able to successfully set up Lee Harvey Oswald and Diana, Princess of Wales on a coffee date. AI is the future. https://t.co/zIt4yK3NMD pic.twitter.com/BXyObc5gEl

— Mutual Aid Art Raffle📌 Paulie ◼️◼️ 551/551 (@paulieponzi) January 19, 2023

i can’t breathe https://t.co/y6ig16nAQE pic.twitter.com/y2nrhjyFLQ

— katie, paper bag princess (@ktholdss) January 19, 2023

A historical mystery SOLVED!! Henry the 8th never got head ever pic.twitter.com/2o8NIFyu5i

— Your Local Dingus (@Dinkalus) January 18, 2023

Nah holdup let this app cook https://t.co/OFoAhPWOOx pic.twitter.com/kY5AqTu7cX

— Faye ☭ (@yelanusssy) January 19, 2023