Filmmakers Probe Early Chatbot Abuse
A new documentary investigates why millions of users insulted the popular early 2000s bot SmarterChild.
By Michael Collins ·

The rise of artificial intelligence has sparked renewed interest in how humans treat digital entities, a theme explored in a new documentary. The film, titled Searching for SmarterChild, recently won the New Visions Award at the Ji.hlava Documentary Film Festival. It examines the history of one of the most popular early chatbots in North America.
Zan Gillies and Lindsey Sitz co-direct the U.S.-Canadian project, which is produced by Erin Williams-Weir. The film focuses on SmarterChild, a bot available to millions of households through AOL, Yahoo, and other platforms two decades ago. Gillies states that approximately 3 million people engaged with the bot via text messages. He notes that these conversations were often markedly rude.
The directors realized the project’s potential two years ago when they discovered they had both used SmarterChild in their youth and treated it abusively. Gillies tracked down the inventors and engineers behind the bot, who retained digital records of every chat session. One of Gillies’ own saved conversations showed him telling the bot his favorite hobby was "Setting things on fire." When the bot suggested an alternate activity, he replied, "Setting YOU on fire."
Gillies says the core question of the film is what our relationship with artificial intelligence reveals about ourselves. He argues that while chat technology has evolved, the fundamental questions about our interactions remain unchanged. He asks whether these machines are friends, therapists, assistants, or lovers. The filmmakers have found that this tendency to insult bots is not unique to North America, with similar behavior observed among users in Korea and other regions.
Silicon Valley engineers were surprised by the level of abuse directed at SmarterChild and held emergency meetings to address the issue. They developed a feature that recognized rude language and responded with warnings and snark. If the insults continued, the bot would enter a temporary shutdown until the user apologized. Gillies was surprised to learn the programmers created a rating system to calculate the relative weight of different insults.
The filmmakers have interviewed members of both the West Coast and East Coast teams that composed the response language for SmarterChild before it was sold to Microsoft and shut down. They have also spoken with Lauren Kunze, CEO of Pandorabots, who researches the parts of the human brain triggered by AI interaction. The final version will include a sequence where a rebuilt SmarterChild bot converses again.
Gillies and Sitz have raised approximately $18,000 on crowdfunding platforms, which they expect to cover about a fifth of the final budget. They plan to interview more experts on human-AI interactions, including a conflict resolution specialist focused on apologies. The documentary aims to capture the full scope of this early digital relationship.
Source: Variety






