Joel Pinneae encountered a concept of artificial intelligence bias a quarter century ago as a student at Wateralu University, Ontario. She was recruited by a university project to support the training of the helicopter pilot’s audio recognition system.
In order to build a fair system, they wanted female pilots to train technology, but they could not find anything. As a result, Pinnow, a graduate of engineering sitting next to the cockpit pilot, was recorded to train systems at a stress level that the pilot usually experienced.
“I accepted the challenge,” she recalls. “It was a very strict science, this experience of calling your data for bias and finding a solution around it. I didn’t have enough samples to actually create a powerful case. However, at least it has a scale of the system and performance for reporting the gap. “
The 49 -year -old Pinnow is actually talking from a New York office room in Meta, named “reproducible research.” She is currently the Vice President of AI researcher at Meta, a high -tech company that owns Facebook, Instagram, and WhatsApp.
Meta is the leader of AI development. The basic AI Research (Fair) LAB aims to conduct long -term research on meta and achieve a scientific breakthrough of technology implemented on the entire platform.
Mark Zuckerberg, the highest executive officer, has outlined his vision of building general information that the system has an intelligence equal to most humans. META wants to build this in open source, and wants to make it easier for developers and scholars to use techniques, not behind a closed system.
Rama, a series of large -scale language models of open source, has closed its models from competitors such as Openai, Google, and Anthropic. It is used in Meta AI. This can be used for interfaces that users can chat on and on Facebook, and images.
META’s advanced research is conducted on Fairs, which leads about 1,000 Pineau at 10 locations. “That’s a big responsibility,” she says. “Sometimes I look at things in the lab, but it’s a few months ago in my hands, even if it’s not billions.”
What was the way to the up -to -hand musicians and mathematics to form the world’s most powerful technology?
She says Pinnow’s appeal to the complex theme has been born out of love for mathematics “from a very young age.” “I remember being fascinated by the basic mathematical concepts such as power laws and logarithms.” The strong interest in music playing piano and viola in the orchestra is finally a hobby. did.
At the age of 19, Pinnow’s first job was working for the Natural Resources of Canada. Later, she worked in the fire room in Otawa in his hometown. She has obtained a robot engineering Ph.D. at the University of Carnegie Melon, where she built a lowing robot that detects obstacles and assistants in nursing home.
At that time, she discovered AI and machine learning and built an algorithm and conversation chatbot. This is an advanced technology primer behind Openai’s Chatgpt bot supported by Meta AI and Microsoft. “It was very primitive at that time in the late 1990s, but it was fun.”
While teaching computer science at McGille University in Montreal, Pinnows say, “It’s very clear that many of the largest innovations of AI will happen in the industry.” Therefore, in 2017, she joined a meta, known as Facebook, and opened a fair lab in Canada, along with existing labs in Paris and New York.
“I flew with curiosity about what I could do in research when there was an industry -level investment,” she says. “Meta was the only (company) that committed to an open research opened, and I didn’t care about the interview elsewhere.”
She said, “It’s a bit difficult to find people who have more appetite to have the ability to conduct research and more responsibilities.” But she is now a member of a female leadership team than a man.
“I didn’t hire most of these people (fairly), but I had the opportunity and supported their roles,” she says. Then she adds. “It’s easy to attract other women as a woman’s leader. It’s another important part of the equation.”
Among her college careers, Pinnow is one of the 15 women in 75 cohorts in Waterlou, one of the six carnegie melon robot engineering. It was. Today, the number of women participating in the high -techescator remains low, and in 2019, the professional service company, Accenture, has left half by the age of 35.
“What I say, you need to pass through the” Survivor Bias “filter. 。 。 Probably because I actually had a very positive experience. She says. Pinnow quotes the support from many generous colleagues and contacts that he had over time. However, she also “recognized a lot” the imbalance between men and women in this field.
Pinnow believes that academia is making an “amazing progress”, and the means such as enabling students to specialize or change the subject during the course can help improve gender balance. There is.
She still belongs to Magil University and lives in Montreal. In Montreal, the California counter part and the time with friends other than the high -tech world, “I give me a very different perspective.”
With the long history of working in AI, the distance gave her precious perspective because the world is working on hype and concerns about the potential of this powerful technology.
Recommendation
“No one can actually predict the future, but there’s definitely a moment (happening),” she claims. “For the first time since the launch of Chatgpt, everyone, every day, anyone can experience AI in a completely different way. Before that. That was invisible.”
There are still many “open research issues” for fair, and more meta has questions about the product and regulations.
“There are many scrutiny, but there are many really good and important questions needed to make a decision as a society,” says Pinnow. “How do we intend to position yourself? There is no correct answer or the wrong answer. How to cooperate in supporting innovation, maintaining people, and creating an environment that respects creators’ work. what?
“We have to find the right way to weave them together.”
