RSM Open Lecture “Challenges and opportunities of AI” featuring professor Ting Li and the guest speaker from Deloitte Stefan van Duin left me pleasantly surprised about the positivity these experts share regarding the future of AI*. While I recommend watching the whole lecture, one idea particularly struck my mind: professor Li predicted the AI career advisors who would help you choose your career path, or at least stir your choice away from the one not suitable for you. Indeed, in the world where social media algorithms can predict your actions better than you can yourself, where better to turn to for help when making such a decision. AIs are already used by some companies to help screen the CVs, so why not optimize your chances with the help of another AI? A quick Google search revealed that such options already exist [see e.g. 1, 2]. And hereunder, I would like to share with you an article that proves that the future is closer than we might have expected.
In 2019, Kern, McCarthy, Chakrabarty & Rizoiu used Twitter posts to analyze how the personality traits relate to people’s professions [3, 4]. First of all, they used the information to create the “digital fingerprints”: determine people’s Big 5 personality traits and 5 other core values. They then crossed this information with their profession stated in their profiles and created a 10-dimensional map according to these 10 traits. For clarity, this model was converted into a 2D space [3] (please find the vocational map below). You can see that these professions have been arranged into clusters according to their resemblance; I highly encourage to follow a link and check out the professions under these clusters (see [5]). Then, they checked how accurately they could predict the profession based on these traits [3] (please find the bar chart below). The accuracy is astonishing, especially taken into account that some of the most prevalent mix-ups refer to the professions very similar to each other, e.g. school principals would be mixed up for teachers (which has indeed been their past profession), or data scientists for software engineers [3] (please find the chart below).
Therefore, based on the personality traits, the AI is actually able to predict the current profession; now, with the first generations growing up with their life fully recorded, we might have enough data to build models to train AIs to predict our future personalities, and consequently the best careers for us. But one problem remains: people are generally not happy with their choice [3]; how to train the AI to lead us to a happy future?
* Note: the recording of the lecture, dated 7 October 2020, might later be posted on the RSM Youtube channel, but sadly is not available at the time of the writing).
[1] https://trustifyai.com/career-robo-advisor/
[2] https://skillroads.com/
[3] https://www.pnas.org/content/116/52/26459
[4] https://theconversation.com/robot-career-advisor-ai-may-soon-be-able-to-analyse-your-tweets-to-match-you-to-a-job-128777
[5] https://www.rizoiu.eu/documents/research/resources/Vocation_Map_Interactive.html
Hi there Agne,
I am glad to read I was not the only one who was pleasantly surprised by some of the AI applications that were named during the open lecture. I too liked the idea of an AI being able to recommend your future career path. In Ting Li’s example, she also named the ability of the AI to recommend the level in which someone could follow a course on the university. This seems very interesting to me. For example: not everyone is as great in fields like math as others, so how useful would it be if an AI can assess your skill level and recommend specific areas to focus on to improve your skillset.
I do agree that there are challenges, with regards to happiness or changing interests. A lot of entrepreneurship comes not from initial fields interest, but rather from unexpected problem-solution situations. I still have a hard time imagining AI predicting Mark Zuckerberg starting Facebook or even Warren Buffet starting a hedgefund. However, AI might direct them to their eventual sector, in which they then can let their entrepreneurship guide them further
I have never really thought about this before. But it does make a lot of sense to train an AI on resume’s. Although of course it is unstructured data. There is quite a lot of structure in a resume and the structure likely also influences your chances of landing your dream job. Especially if there are many applications and the recruiter has to quickly scan through a lot of resumes or has automated part of the process as well. It could prove for some interesting business models as well since this information can have value for many different parties.
Hi Agne! Thanks for sharing such an interesting topic! As a student who is always trying to find a work, I found it very relevant and useful as for myself, it is always difficult to analyse from the job description and company profile whether I suits this position or not. Having an AI career consultant would be very supportive in many aspects! Using “digital fingerprints” could help imply more accurate advice and avoid bias that might occur if people take the personality tests themselves. However, some concerns that raised while I am reading this blog: will privacy be an issue for collecting the data? From the article of “digital fingerprint”, it analyses the traits mostly based on people’s professions, and does not really take psychological factors into account. That might be one of the reason that people are not happy with AI’s career suggestions. For example, I was a top student in biology and chemistry in my high school, AI might suggest me study medicine and pursue a physician career. However, what AI does not know is, I do not even dare to touch a cat…
In the future, I would like to see more advanced and “smart” AI career advisors that are more considerate, allowing better interactions between users and advisors, as well as offering personalized options.
Overall, I am still very excited to see how it will be developed and applied in the upcoming future!