In the previous chapter we understood the nuances involved with Ethics in AI.
Today let's take a look at a new chapter in our Story.
Who is Accountable for what your AI does? Who do you contact to fix your AI problems? Who needs to make sure that AI serves its purpose?
These are some of the questions that we as AI designers many a times overlook. The criticality of such questions plays a crucial role in building trust and reliance between the users and the product. It is possible that your product may be providing state of the art results for the larger community, but if it doesn't solve it purpose even for a small segment of the society the distrust starts building up.
As AI is relatively a newer chapter in the book of Humanity, it is essential to make sure that anything and everything performed by such an evolved system should be as close to PERFECTION as it can get, in this case HUMANS.
Human decisions play a very significant role through out the life cycle of any AI based system. Be it Data Gathering, Feature Engineering, Neural Architectures, Training, Testing, Validation or even Deployment.
A fact that we all need to understand is AI is not perfect. It imitates and attaches itself to the characteristics of its creator and as no human can provide 100% accuracy in the same manner the system fails to do so.
AI is like a child. One has to teach and guide the child through out its childhood till the time it hits maturity and even still it needs to keep learning from the environment. Anytime the child makes a mistake, it's the parents' responsibility to correct the child's actions and punish it for the same so that it doesn't repeat the same.
The analogy used above is the reality of a development life cycle of AI. No matter where, what or how we look at it, it is the responsibility of the Designers or the architects to be able to provide the best and trustable AI to the public. Because it is not just a niche of people who use AI, but it is the whole world at it.
As it was well quoted once by IBM -
Every person involved in the creation of AI at any step is accountable for considering the system’s impact in the world, as are the companies invested in its development.
We as AI designers have some things under control and some we don't. It is very essential to understand where the line needs to be drawn. Some of the guidelines that every Organisation owning or building an AI product should follow are -
- As an AI designer or developer it is your responsibility to know the company policies regarding data management.
- One needs to understand the extent of a company's/product's responsibilities and where they end.
- It's very essential to maintain records of Data Modelling, Versioning, and Gathering strategies.
- Last but not the least one should definitely read and adhere to the IEEE Ethically Aligned Design Document.
Everyday there are thousands of fresh minds pouring into the AI community and contributing significant research to the same. It is the responsibility of the Product Owners to make sure that the young minds have their train of thought aligned with the Ethics involved with the product, if not, on a long run it can create catastrophic aftershocks.
Some questions for all AI designers to ponder upon every time they innovate would be on the lines of -
- Is the AI to be collaborated in a human decision-making process?
- Is it making decisions and evolving as expected?
- Is it able to accept user feedback and improve or becoming biased?
- Is it over stepping its boundaries in terms of data gathering?
- Is it in line with the Ethics involved with AI?
Everything we discussed until now is just a trailer of the implications and change such systems can bring into our world. Hence it is our responsibility to make sure that we do our best in delivering the finest results back to the society.
According to one of the surveys -
"Nearly 50% of the surveyed developers believe that the humans creating AI should be responsible for considering the ramifications of the technology. Not the bosses. Not the middle managers. The coders"
- Mark Wilson,
Fast Company on Stack Overflow’s
Developer Survey Results 2018
It's essential to keep in mind the responsibilities involved in developing AI based systems which directly impact the world on a large scale. As we are the ones accountable for all the actions performed by the system. As much as you may hate it we are the Customer Care for AI.
I hope this article finds you well and encourages you to develop ethically sounds products for the development and upliftment of the society. 😁