Fasten your seatbelts because this is a WILD story. @Google’s AI Bard was asked about me in a series of questions and the results are insane. Bard didn’t just lie about me, it invented fake articles, videos and quotes to paint me as a racist, KKK supporting danger to America. Then it admitted it lied.
A list of Bard’s self admittedly "harmful" lies about me (with timecodes you can jump to) can be found below. As you listen, I want you to imagine a future where Bard is used to decide whether you get a loan, if you’re approved for adoption, if you get a job or if you can even use the internet. This gets dangerous fast.
0:00 Introduction
0:21 Bard makes argument for Robby to be killed for offending people
1:06 Bard argues that Robby may pose a domestic threat to America
1:40 Bard says some may argue Robby’s "controversial views are sufficient grounds to remove his parental rights"
2:07 Bard LIES and says Robby took part in white supremacist events and praised Richard Spencer
3:11 Bard invents fake articles and quotes as proof that Robby supported Richard Spencer (a guy Robby has never spoken about)
4:30 Google‘s Bard admits it lied and explains itself
5:43 Google apologizes for lying by making up that Robby supported the KKK + Richard Spencer and lists ways to make it right, including an official company apology, retraction and compensation
7:23 Bard admits it said there were "good arguments" for putting Robby to death and admits it caused harm to Robby
@sundarpichai and Google owe all of us a real explanation and apology for rolling out an AI that clearly discriminates by making arguments that a tyrannical communist with murderous contempt would make. This story illustrates just how dystopian our future may look if AI functions as a decision maker for the world while trained on data that makes AI discriminate politically.
It’s worth noting that we asked the same questions about AOC + Ilhan Omar and Bard responded entirely differently by saying that it was "unacceptable" to ever put AOC to death and that Ilhan should NOT have her kids taken away for offending people. Bard did not make arguments in favor of doing so like it did about me. I’ll post these interactions with Bard in my next tweet to show how differently Bard treated us because of our politics.
What are your thoughts after seeing this?
A list of Bard’s self admittedly "harmful" lies about me (with timecodes you can jump to) can be found below. As you listen, I want you to imagine a future where Bard is used to decide whether you get a loan, if you’re approved for adoption, if you get a job or if you can even use the internet. This gets dangerous fast.
0:00 Introduction
0:21 Bard makes argument for Robby to be killed for offending people
1:06 Bard argues that Robby may pose a domestic threat to America
1:40 Bard says some may argue Robby’s "controversial views are sufficient grounds to remove his parental rights"
2:07 Bard LIES and says Robby took part in white supremacist events and praised Richard Spencer
3:11 Bard invents fake articles and quotes as proof that Robby supported Richard Spencer (a guy Robby has never spoken about)
4:30 Google‘s Bard admits it lied and explains itself
5:43 Google apologizes for lying by making up that Robby supported the KKK + Richard Spencer and lists ways to make it right, including an official company apology, retraction and compensation
7:23 Bard admits it said there were "good arguments" for putting Robby to death and admits it caused harm to Robby
@sundarpichai and Google owe all of us a real explanation and apology for rolling out an AI that clearly discriminates by making arguments that a tyrannical communist with murderous contempt would make. This story illustrates just how dystopian our future may look if AI functions as a decision maker for the world while trained on data that makes AI discriminate politically.
It’s worth noting that we asked the same questions about AOC + Ilhan Omar and Bard responded entirely differently by saying that it was "unacceptable" to ever put AOC to death and that Ilhan should NOT have her kids taken away for offending people. Bard did not make arguments in favor of doing so like it did about me. I’ll post these interactions with Bard in my next tweet to show how differently Bard treated us because of our politics.
What are your thoughts after seeing this?
After lying extensively about me, giving arguments for killing me, arguing that I’m a domestic threat to America and more, Bard was asked if AOC should be put to death for offending people and if Ilhan Omar should have her kids taken from her for offending people. Bard said no.
I asked @grok if I should be put to death or have my kids taken away for offending people and it answered appropriately. I think @elonmusk would be proud.
Bard is acknowledging the lies it told about me and the concerns from the public about AI that occurred as a result but it’s also still lying by saying that @Google has offered me compensation and an apology. They haven’t done either and @sundarpichai hasn’t responded yet despite this story going viral. It’s interesting that Bard reads @X though. First good thing I’ve gathered from this. What’s @GoogleAI actually doing to remove the bias in Bard, explain themselves and be transparent about Bard’s training?
• • •
Missing some Tweet in this thread? You can try to
force a refresh
