Supported by
Facebook Gives Workers a Chatbot to Appease That Prying Uncle
The “Liam Bot” teaches employees what to say if friends or family ask difficult questions about the company over the holidays.
Sheera Frenkel and
SAN FRANCISCO — Some Facebook employees recently told their managers that they were concerned about answering difficult questions about their workplace from friends and family over the holidays.
What if Mom or Dad accused the social network of destroying democracy? Or what if they said Mark Zuckerberg, Facebook’s chief executive, was collecting their online data at the expense of privacy?
So just before Thanksgiving, Facebook rolled out something to help its workers: a chatbot that would teach them official company answers for dealing with such thorny questions.
If a relative asked how Facebook handled hate speech, for example, the chatbot — which is a simple piece of software that uses artificial intelligence to carry on a conversation — would instruct the employee to answer with these points:
Facebook consults with experts on the matter.
It has hired more moderators to police its content.
It is working on A.I. to spot hate speech.
Regulation is important for addressing the issue.
It would also suggest citing statistics from a Facebook report about how the company enforces its standards.
The answers were put together by Facebook’s public relations department, parroting what company executives have publicly said.
And the chatbot has a name: the “Liam Bot.” (The provenance of the name is unclear.)
“Our employees regularly ask for information to use with friends and family on topics that have been in the news, especially around the holidays,” a Facebook spokeswoman said. “We put this into a chatbot, which we began testing this spring.”
Facebook’s reputation has been shredded by a string of scandals — including how the site spreads disinformation and can be used to meddle in elections — in recent years. In October, Mr. Zuckerberg was grilled by lawmakers for hours over everything from political advertising to work-force diversity.
For Facebook employees, that has sometimes led to questions from family and friends about why they would work for the Silicon Valley company. This year, Facebook fell to seventh place from the top spot when people were asked where they most wanted to work, according to a survey by Glassdoor, the employee reviews site.
Facebook announced the Liam Bot in a message to employees, the company said. In past years, Facebook provided employees with guidance for the holiday season by sharing news releases in internal groups, or directly with those who asked for advice. The company said it wanted to create a more efficient way to answer employee questions this year.
In its answers, the Liam Bot often links to company blog posts and news releases. It doesn’t just provide answers to difficult questions about Facebook’s role in the world, either. Liam Bot is also practical with personal technology advice.
For instance, if relatives asked a Facebook employee what to do if they were locked out of their Facebook accounts because of a password reset, Liam Bot could provide a step-by-step solution.
Erin Griffith contributed reporting.
Sheera Frenkel covers cybersecurity from San Francisco. Previously, she spent over a decade in the Middle East as a foreign correspondent, reporting for BuzzFeed, NPR, The Times of London and McClatchy Newspapers. More about Sheera Frenkel
Mike Isaac is a technology reporter based in San Francisco. He covers Uber, Facebook and Twitter, among other companies. He previously worked for Re/Code, AllThingsD and WIRED, and began his career writing about music for Paste Magazine and Performer. More about Mike Isaac
Explore Our Coverage of Artificial Intelligence
News and Analysis
Amazon said it had added $2.75 billion to its investment in Anthropic, an A.I. start-up that competes with companies like OpenAI and Google.
Gov. Bill Lee of Tennessee signed a bill to prevent the use of A.I. to copy a performer’s voice. It is the first such measure in the United States.
French regulators said Google failed to notify news publishers that it was using their articles to train its A.I. algorithms, part of a wider ruling against the company for its negotiating practices with media outlets.
Apple is in discussions with Google about using Google’s generative A.I. model called Gemini for its next iPhone.
The Age of A.I.
The Caribbean island Anguilla made $32 million last year, more than 10 percent of its G.D.P., from companies registering web addresses that end in .ai.
When it comes to the A.I. that powers chatbots like ChatGPT, China trails the United States. But when it comes to producing the scientists behind a new generation of humanoid technologies, China is pulling ahead. Silicon Valley leaders are lobbying Congress on the dangers of falling behind.
By interacting with data about genes and cells, A.I. models have made some surprising discoveries and are learning what it means to be alive. What could they teach us someday?
Covariant, a robotics start-up, is using the technology behind chatbots to build robots that learn skills much like ChatGPT does.
Advertisement