As a student there are a lot of things constantly twirling my mind. Good university application, career path, job skills. I was drowning in summer course after summer course looking for the right things to spend my time on. Until I came across Teens In AI.
It was my first time ever doing a coding summer course. I was unsure of what to expect. Were they all going to be nerds, what if I don’t fit in? I entered Accenture on the first day and to my surprise there was a room full of excited young students in the exact same position as me.
Initially we were split up into 6 teams each tasked with tackling an issue of our choice from 1 of 3 groups. Healthcare, Environment and Education. This gave all of us the freedom we needed, to code for what mattered to ourselves as well as what mattered full stop.
For the first three days, my team was brainstorming. We had several lectures on design thinking which taught me about the importance of accessibility and practicality. A significant yet overlooked aspect of an application. After conferring it was decided that my team would tackle the issue of mental health, in particular dealing with bullying.
Speakers including Alejandro Saucedo and Julien Cornebise came in to lecture. Issues raised from the talks included biased data.In machine learning an algorithms main aim is to lower its cost function output.
A famous case includes the hypothetical ‘jail AI’ whose role was meant to be deciding whether people went to prison. The problem was that the input data was face recognition based, and we all know how controversial that is nowadays. Based on the training sets, black people happened to be featured prominently and as a result the AI recognised the trend in this set of data. As the AI learned it continued to wrongly judge coloured people since it would yield the best loss results. Whilst the AI was right 80% of the time, its inbuilt bias ignored the wrong 20%.
Luckily this flaw was spotted but it begs the question, what other societal bias is present in AI, and is any of it affecting us today?
As well as speakers, the framework of our entire chatbot couldn’t have come to creation without the help of some brilliant mentors include Caroline Matthews, Nagaraj Sengodan and Cihan Dogan. Building was the main focus onwards from design thinking. Every day onwards was tech grind after tech grind.
Every day I learnt more machine learning. Ranging from linear regression to stochastic gradient descent the course was fulfilling. It was easy to pick up even for a beginner. Besides mathematics, platforms played an essential role. Thanks to the mentors, I was introduced to API’s like azure QnA maker and LUIS. Combining my knowledge of programming, maths and tech I was complete with my project.
Nearing presentation day, I was bombarded with tips and tricks towards speaking and putting forward ideas for potential investors. Thanks to speakers‘ advice and team bonding exercises I was comfortable with adapting to surroundings, create a straight forward pitch deck and learnt how to perform efficiently within a team.
At the end of the whole experience I had created a network of new contacts, explored a new branch of mathematical knowledge and gained skills that I can use for the rest of my life.
My passion for technology was met with passions to help the world and as a result I had learnt so much more from the course than just AI. I advise you join too.