Hello everyone, My name is Kenshin Fujiwara and I am the CEO and founder of HACARUS Inc.
Through this series of blogs, I will discuss a wide range of topics of AI, from the history of AI to practical tips for the successful application of AI projects. I hope my blog posts will help you better understand AI and solve your business issues.
In the last blog, we discussed the current progress and future responses regarding AI ethics. Today, we will discuss the advantages of sparse modeling in the human-in-loop approach.
Training AI with Small Amounts of Data
It is important to remember that AI is only as accurate as the humans training it. If the AI is improperly trained, then it will also make mistakes in its decision-making.
Looking at medical imaging once again, it is virtually impossible for a physician to check all of the tens or hundreds of thousands of medical images. However, for applications where sparse modeling was introduced, it provided highly accurate diagnostic support while training with less than 100 images.
With 100 images, the physician can ensure the quality of each one and the accuracy of the AI model. In other words, humans can be deeply involved in the training of AI.
Sparse modeling is also advantageous when retraining the AI system for future changes or updates. For example, if it is found that the AI is making a wrong decision, humans would need to recheck the data used by the system. Since sparse modeling aI is trained using small amounts of data, this process will be quicker and less costly to carry out.
Sparse Modeling Mimics a Human-like Manner
Sparse modeling can also be thought of as an AI that behaves in a human-like manner. For example, when identifying a human face, conventional deep learning AI analyzes every detail of the face and compares it to a database. Then it determines the probability of a match aiming for 99.999% accuracy.
However, humans don’t necessarily look at these details when we recognize a person’s face. Instead, we only look at the distinctive features such as the eyebrows, cheeks, nose, and mouth.
In a way, the human brain is designed around convenience and doesn’t limit itself by analyzing areas that are unnecessary for judgment. Sparse modeling follows this basic idea to mimic the human thought process. I believe that this is the best way to develop AI for the human-in-loop approach by combining AI with the knowledge and know-how of humans.
Anxiety about the Advancement of AI Technology
Shifting to pop culture, there was a British science fiction film released in 2015 called Ex Machina. The film features a humanoid AI called Eva and a young man who has fallen in love with her. In the end, the man sacrifices himself so that Eva could be freed from humanity.
It is the story of a quiet revolt by AI that has become impossible for humans to control. The film successfully evokes the fear and pessimism about AI that lies dormant in many people.
In reality, we have seen a lot of stories in the media where people are anxious about the advancement of AI technology. A common topic against AI is that it will take away the jobs of hard-working people in the future.
As an AI optimist, when I imagine the future that the Human-in-the-Loop methodology will bring, I cannot help but think that no matter how advanced AI becomes, a bright future where humans and AI will coexist will be waiting for us. Perhaps a future will come in which humans and AI will work together by combining their unique capabilities to accomplish something new.
For now, I think that the idea of AI uprisings, where they take over the world, should be left in the writings of science fiction.
In the next blog, I will discuss the competitiveness of sparse modeling in the business context. For your updated knowledge and insight about AI technology, subscribe to our newsletter or visit HACARUS website https://hacarus.com.