Machine Learning and AI

 

What Machine Learning is and its Relations to AI

What I understand about machine learning is that it is a tool that can learn. The machine is like us humans in that it has its intelligence. However, intelligence is not acquired as we do from engaging in socialization activities or pursuing a degree from school. Instead, machine learning earns its intelligence through programming made by humans. Then, how does machine learning correlate with AI? I read an article here that says that machine learning and AI are linked where machine learning is a subfield of AI. In other words, AI is the umbrella under which machine learning goes. In addition to machine learning, there are other subfields under AI, such as deep learning, natural language processing, and others.

 

The two videos (see the references below) explain some strategies to ensure this technology can reduce and prevent unfairness and/ or bias. For example, some AI tools need to be able to identify faces correctly due to incomplete training of the AI developers. Buolamwini (2016) said that we should speak up when users face this issue. When we speak up, our experience can be the source of how this type of technology should behave, and we will not repeat the same mistake. We might also want to join the Incoding movement (explore it here) to fight the bias issue caused by AI. 

In addition to that, another strategy, as stated by Sharma (2018), is to be aware of human biases. After all, this machine is built by humans, and without realizing it, we put ‘us’ in it. Therefore, when creating this machine, we must set aside biases. Second, she further said that a team of people from different backgrounds should be considered when building the machine. I am on the same page with her as diverse team members will bring diverse backgrounds such as gender, race, and culture, which also leads to checks and balances from the team members to avoid any biases. Finally, Sharma (2018) said that the machine needs more exposure to learn to avoid the issues. For example, instead of learning to recognize the faces of one particular race, the machine should be able to detect all races (the experience of Buolamwini). This way ensures that fairness is upheld and supported. 

You can also explore AI vs Machine Learning more through this video. 

References 

Buolamwini, J. (2016, November). How I'm fighting bias in algorithms [Video]. TEDxBeaconStreet. https://www.ted.com/talks/joy_buolamwini_how_i_m_fighting_bias_in_algorithms?subtitle=en

Sharma, K. (2018, March). How to keep human bias out of AI [Video]. TEDxWarwick. https://www.ted.com/talks/kriti_sharma_how_to_keep_human_bias_out_of_ai?subtitle=en