These five women are changing the AI industry

Written by: Li Yuan

Edited by: Zheng Xuan

Source: Geek Park

Unfortunately, media resources in this world are not always distributed evenly according to importance.

Similarly, in advancing the frontiers of technology, Demis Hassabis, the leader of DeepMind who developed AlphaGo, which defeated human chess players, and AlphaFold, which may potentially benefit humanity immensely, is not as famous as Sam Altman, the leader of OpenAI, who staged a palace intrigue.

Combined, they might not be as well-known as Elon Musk, who attempted to invest in both of their companies and failed.

Musk is undoubtedly outstanding. However, the media resources he receives far exceed those of equally outstanding individuals, and news related to him, even if not of high importance, often makes headlines—before Musk took over Twitter, I really don’t remember seeing "Twitter might add a small feature!" frequently in tech news, and it was just a possibility!

Musk dares to act and speak. For every internet celebrity entrepreneur like Musk, there are always ten low-profile promoters like Demis Hassabis, who enjoy a reputation in the industry but are not well-known in the public domain.

When it comes to women, this situation is often even more severe. Female tech workers often carry a double debuff; on one hand, they tend to be low-key, and on the other hand, they face structural disadvantages from society.

When doing the same things, women's scientific achievements are often overlooked and attributed to their male colleagues working alongside them—this phenomenon has long been recognized and named the Matilda Effect.

For instance, when thinking about programming and artificial intelligence, how many people can immediately recognize that the first computer program in human history was written by Ada Lovelace, a woman, and that the first textbook in AI came from Elaine Rich, also a woman? Among those who have long followed the AI field, how many can name a few outstanding women in AI?

It's okay; before doing this topic, I couldn't immediately name many such outstanding women either, but that doesn't mean there aren't remarkable female tech workers engaged in the AI industry. This is precisely the significance of International Women's Day.

On March 8, let's take a few minutes to learn about five outstanding female AI researchers and entrepreneurs.

01 Fei-Fei Li, Creator of ImageNet, Sparking the AI Image Explosion

The explosion in the AI field has a coherent history in academia, and in industry, it can almost be traced back to one moment: in 2012, the deep learning network AlexNet achieved a very high success rate in image recognition.

Since then, artificial intelligence has gradually entered an era dominated by deep learning, and within a decade, AI has become a buzzword in all our lives.

The proposal of AlexNet ultimately traces back to the ImageNet established by Fei-Fei Li in 2009.

Fei-Fei Li was born in Beijing in 1976 and grew up in Chengdu. At the age of 12, she moved to the United States. At that time, she could hardly speak English, but within two years, she quickly achieved a strong command of the language while also demonstrating strong mathematical abilities. In 1995, she entered Princeton University on a scholarship, and she would return home almost every weekend to help her family manage a dry cleaning business they had opened with borrowed money.

In 2007, Fei-Fei Li became an assistant professor at Princeton University. At that time, researchers in the field of computer vision typically needed to write a specific algorithm to recognize dogs and another to recognize cats.

Fei-Fei Li's intuition was that the model's capability might be sufficient; the problem lay in the data.

Fei-Fei Li | Image Source: YouTube Channel National Geographic Society

She wanted to create a massive database that labeled every possible object in each image. At that time, such projects were almost ignored.

She first had Princeton students work part-time to build ImageNet, but progress was slow. Later, she used crowdsourcing platforms to have part-timers from around the world help with data labeling.

"The online workers, their goal is to make money in the simplest way, right?" she said in an interview with Wired. If you ask them to select pandas from 100 images, how can you prevent them from clicking randomly? Therefore, she embedded and tracked some images, such as photos of golden retrievers already correctly identified as dogs, as a control group. If the crowdsourced workers could correctly label these images, it could be assumed they were working honestly.

The ImageNet project she initiated initially collected 3.2 million images, later increasing to 15 million. It was on this database that researchers could compare whose algorithms were more powerful. And AlexNet, in 2012, rose to fame in the ImageNet competition.

It can be said that ImageNet paved the way for advancements in deep learning, with fields such as autonomous vehicles, facial recognition, and object recognition all starting from ImageNet.

Even today, when people mention breakthroughs in data in certain fields of artificial intelligence, they often use the phrase, "Is this its ImageNet moment?" to describe it.

In recent years, in addition to continuing her research work, Fei-Fei Li has also focused on increasing diversity and inclusivity in artificial intelligence, advocating for resources for the AI academic community so that it does not fall behind the industry.

In 2023, her book "The World as I See It: Curiosity, Exploration, and Discovery at the Dawn of the AI Era" was published, telling the scientific stories she has personally experienced and her interpretations of significant historical moments in AI this century.

02 Niki Parmar, One of the Eight Authors of the Transformer Architecture

The wave of large models came into the public eye, perhaps after the emergence of ChatGPT, but the origin of the large model wave undoubtedly comes from the 2017 paper "Attention is All You Need," written by eight engineers from Google.

This paper proposed the groundbreaking Transformer architecture, on which nearly all leading AI companies, including OpenAI's ChatGPT, are built.

I don't know about the readers, but I was once misled by the media's "Transformer Eight" into thinking that all the authors were male.

That is not the case; the third author of the Transformer paper, Niki Parmar, is a female researcher.

Niki Parmar Interview | Image Source: YouTube Channel IIT Bayarea

Niki Parmar is from India and studied at the College of Computer Technology in Pune, India, before coming to the University of Southern California in 2013 to pursue a master's degree in computer science.

Niki became interested in machine learning during her undergraduate years: "I took the MOOCs on ML and AI taught by Andrew Ng and Peter Norvig, and I was curious about the combined power of data, pattern matching, and optimization," she mentioned in an interview.

After graduating in 2015, she joined Google's research division and began to focus on pure research. By 2017, she became one of the core authors of the Transformer.

Regarding research, she stated, "At first, the vast amount of information and research around me overwhelmed me. Focusing on a specific problem and exploring it with peers can help you ask the right questions."

Niki Parmar and fellow Indian-origin Ashish Vaswani, who is also the first author of the Transformer paper, co-founded two companies, Adept AI and Essential AI. She currently mainly manages the latter.

Essential AI secured $56.5 million in new funding last year from tech giants AMD, Google, and Nvidia. Adept AI previously raised $350 million.

03 Daniela Amodei, Co-Founder of the World's Second-Largest Model Company Anthropic

A few days ago, Anthropic's model claimed to surpass the capabilities of OpenAI's GPT-4, which indeed made headlines.

Reports about Anthropic usually mention that it was founded by seven researchers who resigned from OpenAI or that its CEO comes from OpenAI, often downplaying Daniela Amodei—president of Anthropic and one of its two co-founders.

In fact, Anthropic was co-founded by Daniela Amodei and Dario Amodei, who are siblings. The new large model released by Anthropic has been publicly presented by Daniela in many television media interviews.

When promoting Anthropic, it is often mentioned that it focuses more on aligning AI systems with human values than OpenAI, and Daniela Amodei was previously the Vice President of Safety and Policy at OpenAI.

Daniela Interview | Image Source: YouTube Notion Channel

Daniela is of Italian descent and grew up in San Francisco.

Her work experience is relatively diverse. In college, she earned bachelor's degrees in English literature, politics, and music literature. Her early work was more in the political and non-profit sectors, where she developed strong management skills.

In 2013, she chose to join Stripe, which had just been founded in 2010—at that time, Stripe was still a small company, and now its valuation has reached $50 billion, peaking above SpaceX.

Starting from Stripe, she began applying her management and risk control skills to tech companies.

At Stripe, she was responsible not only for team recruitment but also for one of the most critical aspects of payment companies—risk management. She collaborated cross-functionally with machine learning, data science, engineering, legal, finance, and vendor management departments, leading three teams of 26 people to analyze over 7,000 potential fraud, credit, and policy violation cases, achieving a 72% reduction in loss rates from peak levels, reaching the company's historical lowest level.

In 2018, she once again demonstrated her strong strategic vision by joining OpenAI, directly leading two technical teams: OpenAI's natural language processing and music generation teams, while also managing the technical safety team.

In addition to these roles, she served as Vice President of Human Resources, overseeing recruitment, HR planning, DEI, learning and development, and incubating new business operations teams, making her a true multi-talented individual.

In 2021, she co-founded Anthropic with Dario Amodei.

04 Mira Murati, CTO of OpenAI

Although OpenAI is world-renowned, many may not know that its current CTO is a woman, Mira Murati.

Mira Murati joined OpenAI in 2018, was promoted to Senior Vice President responsible for research, product, and partnerships in 2020, and became Chief Technology Officer in 2022, participating in the development of several projects, including ChatGPT, DALL-E, and GPT-4.

During the palace intrigue at OpenAI, she was briefly nominated as the next CEO of OpenAI.

Mira Murati was born in 1988 in Albania and attended high school in Canada.

Her professional background is in engineering; while studying engineering at Dartmouth College, she directly built a hybrid race car as part of a school project.

After a brief stint in the aerospace field, Mira joined Tesla as a Senior Product Manager for Model X, where her interest in artificial intelligence deepened through Autopilot.

Her interest in research is evident; in an interview, she once mentioned, "Boredom is a powerful motivator for pursuing and exploring the frontiers of anything."

Interview with Mira Murati | Image Source: YouTube Channel The Economic Times

The most important project at OpenAI, ChatGPT, is led by Mira Murati. She has also been deeply involved in many significant milestones for the company.

In 2023, Microsoft CEO Satya Nadella invested $13 billion in OpenAI through a significant partnership managed by Murati and publicly stated that she "demonstrated the ability to assemble a team with technical expertise, business acumen, and a deep understanding of the importance of the AI mission."

The latest news on March 8 indicates that during the incident of Sam Altman's ousting from OpenAI, both she and Ilya Sutskever expressed concerns about Sam Altman, which had a significant impact on the final decision. Unlike Ilya Sutskever, it seems that she has not been marginalized within OpenAI.

These public pieces of information certainly do not represent all the facts, but looking at her, who can say that women cannot do technology or politics?

05 Timnit Gebru, Who Upended Google's AI Ethics Team

Recently, Google's model withdrew its text-to-image model due to AI ethics issues.

This reminds me of a major incident involving Google's AI ethics team in 2020.

In 2020, an AI ethics researcher at Google, Timnit Gebru, publicly stated that she was fired. And the reason for her dismissal?—it was precisely because she criticized the biases present in large language models.

Timnit Gebru | Image Source: YouTube Channel Vice News

Timnit Gebru was born in 1983 in Eritrea and Ethiopia, and in 2014, she obtained her Ph.D. in electrical engineering from Stanford University, studying computer vision and machine learning.

After graduating, she dedicated herself to researching issues related to fairness, accountability, transparency, and ethics in artificial intelligence. She is known for a groundbreaking paper co-authored with others, which showed that facial recognition was less accurate in identifying women and people of color, meaning that using such AI technology could ultimately lead to discrimination; her research eventually led Amazon to change its policies.

In 2020, Gebru co-authored a paper with another researcher criticizing large language models and the environmental impact of training them. The paper also raised concerns about the lack of diversity and ethical considerations in AI technology development.

The article was supposed to be published the following year, but Google's AI head Jeff Dean told colleagues in an internal email (which he later posted online) that the paper "did not meet our publication standards." While arguing with the company, Gebru found that her company email was cut off during her vacation.

This caused a stir at the time. Many prominent researchers, civil rights leaders, and Gebru's colleagues at Google AI publicly defended her on Twitter. A petition supporting her received signatures from over 1,500 Google employees, more than 2,000 scholars, nonprofit leaders, and industry peers.

However, in the end, Timnit Gebru still left Google. After leaving, she announced the establishment of an independent AI research institute—"Distributed AI Research," DAIR aims to combat the pervasive influence of large tech companies in AI research, development, and deployment.

As a true warrior, she once stated, "I cannot wait for big tech companies to finally solve the problems brought by AI."

06 Conclusion

A basic fact is this: even with so many outstanding women, the tech and AI circles are still male-dominated.

Changing this involves too much: the pressures women face in academia, the unequal treatment they receive in the investment community, and even the support for women's mathematical education and workplace measures from a young age.

An article cannot solve these problems.

This is also why International Women's Day and many incentive programs for women still exist. In fact, the last woman mentioned in the article, Timnit Gebru, is a disciple of the first woman researcher mentioned, Fei-Fei Li. Sometimes it can be a beautiful cycle.

At the same time, we can still draw strength from these inspiring women on this special day. In this era where media resources are too scarce for them, take a moment on this day to remember them.

Give credit when it's due.