Create Value

In today’s hot data science and AI world that many are in pursuit of, with countless projects, fancier models, more sophisticated networks, beautiful theories, advanced tools and smart people doing it, how should we view, or can we even quantify and measure the projects and models which are already quantitative heavy, so we can have a clearer mind and keep the right focus moving forward?

You can answer: sure, don’t we already have precision, recall, F1 score, AUC, BLEU score, loss function, error rate, cross-validation, generalizability, transfer learning, domain adaption, desire for AGI…..?

But I think there is another answer which is even simpler, and I believe that is whether or not it “CREATE VALUE”. Create value means creating tools and solving pain points among people, better people’s lives, or reduce human errors and accidents. Either that’s creating a NLP system or Vision system help the doctors better read and judge, reduce error rate and provide better care to patients, or creating speech and writing assistant that helps disabled people or children to communicate and learn quicker. It can also be helping people better and quicker search and discover the information/product/services they are looking for, which they wouldn’t otherwise know how to, or detecting fraud and danger to protect people’s financial security, general security and prevent loss. You name it. They all create tremendous value.

Do we create or at least add value, with our current projects, to solve the actual problem and meet the real need so that people can eventually use the tools we created? Do we keep focusing on answering this question when we decide what to work on or how to work? Before we create cars, we know there is a need for faster transportation and people will eventually use cars. 

I try to keep a mindset of creating value and meeting people’s real need which may be even hidden, when working, and I hope you think so and do it too.