Prevent Bias from Creeping into Your AI. If your organization or team is experimenting with AI, you need to be asking yourself an essential question: How can we ensure that the dataset we use to train AI models is representative and doesn't include harmful biases? First, consider the size of your dataset and understand the tradeoffs. Big datasets can train AI to be more effective at certain tasks, such as keeping a human-seeming conversation going. But if a model is too large, it's nearly impossible to rid it of potential biases. Second, assemble diverse teams—including members of underrepresented groups—to collect and produce the data used to train your model. Your goal is to ensure that people with a variety of perspectives and identities are represented and that they have a chance to identify biases or oversights in the data. AI will only be trustworthy once it works equitably—and that will only happen if you prioritize diversifying the data and development teams that make the technology possible. |
|
Today’s Tip |
Prevent Bias from Creeping into Your AI |
If your organization or team is experimenting with AI, you need to be asking yourself an essential question: How can we ensure that the dataset we use to train AI models is representative and doesn't include harmful biases? First, consider the size of your dataset and understand the tradeoffs. Big datasets can train AI to be more effective at certain tasks, such as keeping a human-seeming conversation going. But if a model is too large, it's nearly impossible to rid it of potential biases. Second, assemble diverse teams—including members of underrepresented groups—to collect and produce the data used to train your model. Your goal is to ensure that people with a variety of perspectives and identities are represented and that they have a chance to identify biases or oversights in the data. AI will only be trustworthy once it works equitably—and that will only happen if you prioritize diversifying the data and development teams that make the technology possible. |
This tip is adapted from “8 Questions About Using AI Responsibly, Answered,” by Tsedal Neeley |
|
|
|
|
|
No comments:
Post a Comment