International Women’s Day Personal Story: Why I would tell my younger self to choose a career in AI
There has been much discussion about the lack of women (and overall diversity) in the world of technology. This is especially pronounced in the world of AI. As today is International Women's Day, I thought I would share my story and rationale as to why I believe a career in tech, specifically in AI, is worth pursuing.
My Dall-E generated self #EmbracingEquity for International Women’s Day 2023.
How I got started in AI
I came relatively late to the world of AI. I only really learnt about AI in 2019 when I was somewhat shamed into learning about AI by a non-technical friend. She was helping train an AI model on fraudulent financial transactions, and was aghast that someone who was a career technologist seemed to know less about AI than she did.
It's not like I didn't try and learn about AI before 2019. I started the way I thought everyone did - by asking Google, which promptly directed me to some very useful Python libraries. Though I know how to code in Python, trying to learn about AI by building a machine learning solution using scikit-learn did not work for me. At all. In fact, I ended up putting AI into the "too hard" basket for a number of years.
Back to 2019 - I ended up learning about AI by taking Andrew Ng's AI for Everyone course after having started (and abandoned) other, more technical, courses. Once I got over the initial learning curve, and got what makes AI both artificial and intelligent*, I found I could go back to some of the more technical courses and tools and understand them much MUCH better.
I do sometimes kick myself for not having persisted with learning about AI back then**.
In any case, I then started to wonder how many people had faced similar “learning difficulties” and were being left behind when it came to AI. Especially those who didn't come from a technical background. I started to question this so much that I decided to pursue a doctorate degree in making AI more understandable and relatable to everyone, which then led to me to start AI Shophouse and writing articles like this one :-)
Why do we need more diverse voices in the world of AI?
We need to ensure that AI solutions are unbiased and fair before they can be used. Bias is pervasive at each stage of the AI lifecycle, from bias in AI algorithms and labelling of data, to a lack of representative data, to AI developer biases manifesting in the solutions themselves. While this may seem daunting, there is actually a solution - to disparity test the AI solutions, to ensure that the results achieved do not disadvantage any of the protected groups (race, gender, religion, and so forth). If it isn’t possible to ensure that an AI doesn’t introduce disparities, e.g. if it performs poorly on women, then we should not purchase or use it until it is ready to be used by everyone.
We also need AI advocates who can ensure that user rights, particularly those of minorities, are being represented. These rights include having agency about what AI is being used for, being able to opt out of AI solutions if their accuracy levels are insufficient, mandate for disclosure about when AI is being used, requiring accountability for decisions being made by AI, and demanding plain-English explanations behind the decisions being made by the AI, i.e. if the algorithm cannot be satisfactorily explained, then it cannot be used on anything except low-stakes AI. This advocacy is particularly important to ensure that AI is being used ethically and responsibly, and not purely for commercial gain. It should also increase in importance as AI regulations are introduced***.
What opportunities are there in AI?
In the AI world, there historically has been a focus on technical implementation roles, such as machine learning engineers and data scientists, but there are many AI-related roles beyond that. As we build more people-focused solutions, there will be a focus on ramping up on AI product managers and designers, AI black box testers (such as prompt engineers**** and disparity testers*****), AI educators, AI ethicists, AI regulators and governance. Many of these roles are yet to evolve, and you can work in AI without coding - coding and implementation are only part of the AI ecosystem. This is the same way a tech company employs many people besides software engineers. As awareness and widespread literacy increases, there will also be an increasing need for AI advocates to work with the AI ethicists, regulators and governance folks.
Besides the opportunities to build more people-focused AI solutions, there are innumerable jobs and industries that are going to be transformed by the use of AI******. This is not limited to areas that are technically focused. For example, jobs in sales are already being transformed by the use of ChatGPT to create personalised content and generate leads. The key takeaway from this is that you have the opportunity to transform your job through your adoption of AI, even if it’s “just” starting with using ChatGPT to generate content or analyse and summarise articles.
Conclusion
It may feel intimidating to get into the world of AI, but it is definitely possible. As the world of AI evolves, the “barriers to entry” to AI are decreasing, with AI solutions getting more and more accessible (think: ChatGPT).
Things you can say to sound smart about AI
I believe everyone is capable of learning enough about AI to make informed decisions about why and how it is used.
We absolutely have the right to advocate for fair and responsible use of AI.
AI is going to impact absolutely every aspect of our work and life, so it is essential that we have visibility of when it is being used.
We live in very exciting times for AI, and I think we should make the most of the opportunity.
Footnotes
*Here is one definition of AI that really resonated with me “Artificial intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions.” Read more here.
**This is one thing regret I have - not getting started earlier. Hence why I encourage others to start now.
***AI regulation is woefully lagging behind the technology. Read here for the New York Times’ March 2023 perspective.
****Prompt engineering is a legitimate form of testing AI systems. Though there are some discussions about its longevity as a field, the need to perform black-box testing on the AI will not.
*****AI disparity testing is a field that is in desperate need of resource, as evidenced by the fact that AI solutions are being deployed and then promptly found not to work on key user groups, such as women.
******Which jobs and industries aren’t going to be impacted on some way. If you can think of one - let me know!
