When technology can chat with you, it’s easy to see it as human. But calling a chatbot a thinking, speaking robot gives it too much credit. This can lead people, even judges in legal cases, to miss the human choices behind the words a chatbot produces. If these outputs lacked First Amendment protection, governments could potentially ban chatbots that criticize or disagree with official viewpoints.
Chatbot outputs can reflect the creative choices of both developers and users. They also touch upon a user’s right to receive information. That’s why organizations like the Electronic Frontier Foundation and the Center for Democracy and Technology filed a legal brief in a case called Garcia v. Character Technologies. Their aim is to explain how these complex language models work and what free speech rights are involved.
Human Expression Drives Chatbot Responses
The extent of free speech protection for chatbot outputs is a key question in a recent court case.
- Chatbot output isn’t solely the product of a machine; it reflects human creative decisions.
- These decisions span from selecting training data to creating system instructions for the chatbot.
- Users’ input also shapes what a chatbot says, linking it to human expression.
- Regulating chatbot content must carefully consider free speech rights for both creators and users.
The brief explores how developers’ choices significantly shape a chatbot’s output. This happens during training, where humans might be directed to reward responses aligning with scientific consensus on climate change, for instance, and penalize denial. This chain of human decisions starts with choosing training data and extends to writing system prompts. User instructions also play a role. So, rather than being the speech of a robot, chatbot output often showcases human expression deserving First Amendment protection.
Furthermore, the right to receive information is protected, even if the speaker wouldn’t normally have the right to say it. Users have a right to access the information that chatbots provide.
This doesn’t mean chatbots are beyond regulation or that their harms can’t be addressed. However, the First Amendment requires that any regulations be carefully crafted to avoid unfairly limiting the right to express oneself or receive information through these tools.
The organizations involved hope their brief will assist the court as the case moves forward. The judge has currently decided not to send the specific question for immediate appeal.
You can read their full brief for more details.
