Hummon moved her way up, working at the energy company Tendril before landing the CTO role at Utilidata, a position she’s held since 2018. She received a PhD in applied physics from Harvard and said that as she made the transition from grad school into the engineering field, she noticed a significant drop in the number of women working in the field. “I’ve often felt the pressure of being held to a higher standard in the workplace compared to my male peers,” she said.
Hummon said now, as a leader, she tries to lead by example and create opportunities for women to lead and be recognized. At the same time, men need to be responsible for creating space for more women to enter and stay in the field, she said.
“The change we need is not just in practice; it must be systemic and widespread,” she continued. “It starts with recruiting and hiring, continues with mentoring and coaching, and culminates with fair and equitable recognition and promotion — all while creating a safe and inclusive workplace.”
Her advice to women entering the AI field is to always remember that being a woman can be an advantage. “It’s given you a perspective that’s different from your male peers, and breakthroughs in technology always come from unique perspectives.”
She says to seek opportunities with companies that have shown their commitment to diversity and that have leaders who have showcased their efforts to support women and minorities. “Judge a potential employer by the way they show up in the interview process, not by the statements or the reports they put on their website.”
Building ethical AI
Hummon said that users should know that AI is not something that can solve all problems but rather is an expert assistant to enhance operations, improve effectiveness, and help support.
“Recognizing its limitations and ensuring there are proper checkpoints is key,” she said.
She said it’s always important to understand how generative AI was trained and built in order to understand any potential biases it could have. “This knowledge is a powerful tool when working with language models for data analysis and evaluating the feasibility of solutions,” she continued. “AI is only as good as the data and information it was trained on.”
She also gave some tips on how to build safe AI, explaining that Utilidata made an early decision to keep all of its data local. “Approaching model building using distributed AI computing reduces the amount and frequency of information being transmitted to the cloud, and therefore reduces the chances of a security or privacy breach.”
As for investors, she said — like all the experts we spoke to — they need to evaluate how a company wants to use AI, especially since responsible AI looks different in every industry. “The investment in responsible AI should be commensurate with the risk and complexity of any given company, not an across-the-board standard.”
