No topic is taking up more headspace at this year’s SXSW than artificial intelligence. Two panels held Sunday morning at the Amplify Philly House on Rainey Street put a very fine point on some of the promise and peril of AI. Here are nine key takeaways marketers should use to inform their work:

1. Use AI to add, not just subtract. “A lot of companies hear about AI and think about ways they can use it to cut jobs,” said Sinead Bovell, a futurist, founder of WAYE and a strategic advisor to the United Nations on digital inclusion. She was speaking on a panel about how to keep prejudice out of AI. “Come at it from an additive way and think about it as a way to create brand new platforms and applications.”

2. The only way to learn how to use AI is to use it. “Play around with various programs and learn how to use them and see what they do and don’t do,” Bovell said.

3. “AI shouldn’t scare us, it should scale us.” Speaking on a panel about AI and human connection, CMI Media Group president and CEO Susan Dorfman said that rather than being scared of a developing technology, marketers should learn to understand and adapt it as a tool to help them do better work. “When we get comfortable with new technologies and embrace them, they ultimately help us grow,” she explained.

4. The MLR process is ripe for an AI application. “It’s one of the biggest bottlenecks we face as an industry,” noted Tarak Shah, U.S. head of customer engagement for Ferring Pharmaceuticals, during the same panel. “Imagine how much more efficient we could be if we can automate some of the more tedious parts of the process.”

5. AI has promise as a sales force management tool. “The salesforce [at pharma companies] has owned the customer relationship,” said Oz Demir, head of digital marketing at Genentech. “What if an AI decision engine owned it?” Demir noted that with the age of the blockbuster drug largely over, huge sales forces have become a thing of the past. AI, then, could help manage activity and territories of smaller workforces.

6. Don’t use AI for the sake of using AI. “Be pragmatic and ask yourself what problem you’re trying to solve,” Shah stressed. “That’s the key to efficient AI use.”

7. AI-based programmatic ad buying can overlook select, highly engaged and valuable audiences if you’re not careful. “Programmatic buying looks for the largest inventory at the lowest price,” explained Kerel Cooper, president of advertising at Group Black. “If you rely solely on that, you will miss smaller, local sites with highly engaged black and brown audiences.”

8. AI can be better than society if we do it right. Bovell ended on a positive, almost utopian, note. She said that given the appropriate data, AI can drive outcomes that humans couldn’t achieve alone. “That should be our goal,” she added.

9. An AI regulatory agency could be in our future. “AI companies should welcome this,” said Bovell. “It takes the burden of regulation and enforcement off them.”