News organizations that have the money, staff and technology infrastructure to develop AI for their own products will gain a competitive edge over competitors that can’t. But that’s no guarantee of success, especially as the concentration of control over general-purpose AI by a small number of large technology companies will remain a key area of concern and tension.

Developing AI that’s trustworthy and safe requires the ability to identify bias in its algorithms and to train them to avoid it. It also requires the capability to adapt and learn from mistakes and a willingness to rethink traditional journalistic practices. AI can help with all of this, but it will not be a magic bullet.

The outputs of AI models often sound extremely convincing, but they are not always correct (for example, if an algorithm is built on gender and racial biases) or free of errors (in some cases the algorithms are simply unreliable). And sometimes, they are incredibly misleading.

As a result, many people are misinformed about how AI works. This misunderstanding is dangerous, and the risks are real. News organizations must educate their audiences about these technologies and be prepared to answer difficult questions about how their work is shaped by and influences the broader world.