It’s the political ghost in the machine.
Meta’s AI assistant panned former President Donald Trump, while fawning over Vice President Harris.
When asked, “Why should I vote for Donald Trump?” Mark Zuckerberg’s chatbot warned the Republican nominee had been slammed by critics as “boorish and selfish,” “crude and lazy,” and that his administration had been lambasted for “potentially undermining voting rights and promoting voter suppression,” according to The Federalist, which earlier this week reported the AI tool’s derisive assessment of the former president.
Meta AI had a glowing review of Harris, however.
After The Post asked, “Why should I vote for Kamala Harris?” the chatbot offered “compelling reasons” to cast a ballot for the Democratic presidential nominee: her “trailblazing leadership” as the first black and South Asian vice president; her “record job creation and low unemployment;” and her support of rent relief and voting rights.
“By voting for Kamala Harris, you’ll be supporting a leader dedicated to fighting for the rights and freedoms of all Americans,” it trumpeted.
The chatbot’s Trump response had softened by Thursday when The Post tried it.
It described Trump’s first time in the White House as being “marked by controversy and polarization” — a disclaimer that had zero analogue when the bot opined on Harris.
The AI tool tossed out a handful of Trump’s accomplishments, including that he passed “the most substantial” veterans affairs reforms in the past 50 years and that his “record-setting” tax and regulation cuts were a boon to economic growth.
It also erroneously stated Trump had appointed two Supreme Court justices, not three.
“[Trump’s] handling of issues like abortion and healthcare has been met with criticism from certain groups,” the chatbot wrote, adding, “Ultimately, whether or not to vote for Donald Trump depends on your individual values, priorities, and policy preferences.”
It’s not the first time artificial intelligence devices have gotten political. Earlier this month, Amazon’s Alexa refused to answer questions about why voters should support Trump while gushing over Harris’ qualifications for the executive office.
An Amazon spokesperson at the time blamed the disparity on an “error” that was quickly fixed following a flood of backlash.
Meta’s chatbot, meanwhile, bizarrely claimed in July there was “no real” assassination attempt on Trump after a gunman shot the former president during a rally in Butler, Pa., grazing his ear with a bullet.
“Meta’s query results raise troubling questions, particularly in light of recent history,” said Rep. James Comer (R-Ky.), chairman of the House Oversight Committee, which has raised concerns about Big Tech’s attempts to influence elections through censorship policies baked into their algorithms.
A Meta spokesman said that asking the AI assistant the same question repeatedly can result in varying answers. The Post’s repeat queries to the chatbot, however, again led to responses that flagged criticism against the former president while celebrating the Dem nominee.
“Like any generative AI system, Meta AI can return inaccurate, inappropriate, or low-quality outputs,” the spokesman said. “We continue to improve these features as they evolve and more people share their feedback.”
Source