YouTube videos watched by engineers I'm Takumi Yobinori an educational Youtuber I normally make lesson videos for college students Most of those lessons are on science topics Math and physics are the main subjects of my videos I believe we are receiving comments from engineers who use AI in the field That's definitely the sense I get from the comments So, I get that they are watching regularly Of course, I have dealt with the subject of what machine learning is and what deep learning is in general and these videos get a lot of views and are seen by a lot of people What these engineers actually watch focus more on specific points of mathematics that they don't understand That's the impression that I get About the 'Shogi and AI' video Machine learning and AI technologies have been around for a lot longer than most people think I think one of the big changes lately is that we're just starting to see them around us more That kind of AI technology is becoming more accessible Say, for example, in my favorite field Shogi - I like board games, especially Shogi I watch many different broadcasted games where a numerical value called "rating value" shows who is winning in terms of AI With more things like that, the general public is now more familiar with AI I think it makes you wonder about the logic behind these numbers There's a video that's been viewed by a lot of people of a complete guide on the history of Shogi AI It's a video that's probably close to an hour and a half long From Shogi AI's first stages to the most recent level, which is said to have surpassed humans I explain the entire history of Shogi AI More people viewed it than I had imagined, so I was really happy but by talking about the types of challenges faced and explaining the technological innovations made at that time I was able to keep people interested at all times Machine learning and mathematics AI itself is a very broad term so it's not so much machine learning as it is the process of AI learning and within it, numbers have a purpose and I need to extract that from the data and output something in line with that purpose In terms of how far the output is from the correct answer and if it's far from the correct answer, how it should be improved it's all determined by mathematics Differentiation, for example, which is taught in high school math is a quantity that shows how much something changes when you change it a little In the case of the data I just mentioned when the actual correct answer deviates from the output and you decide to adjust it by a certain amount the rate of differentiation can be used In that sense, mathematics is really commonly used in machine learning and statisticsis the process of taking a large amount of numerical data and extracting what is meaningful to us For example, statistics that we're most familiar with in our daily lives would be averages Let's say, if we were given the data of the height of all ninth graders in a middle school we would look at it and not be able to understand anything but if you look at one mean value, you can see some trends in the data In that sense, statistics is a process of extracting something meaningful from a large number of numerical data Machine learning or data science is the process of taking large amounts of data and finding a way to bring large amounts of data together to make it useful and statistics is used in the process What is data?based on my research background is food - that's my interpretation of data Why? Because the simplest way to think about it is that AI feeds off of it to learn In that sense, it could be food for AI The research I was doing, information thermodynamics is another theory, so it's important to know what it's referring to I studied this theory in biology, specifically the chemotaxis of E. coli which is the transmission of information that E. coli collects from the external environment and transmits it inside the body The collected information might be, for example the concentration of nutrition for E. coli So it collects these types of information and decides how to move forward That's how I got the idea of how information can be seen as food Furthermore, what is critical about food is that humans cannot get nutrition just by eating food but how we digest food is also important I believe that digestion is like the process of analyzing data using statistics and making it useful So if you ask me what data is my understanding is that it is food before it gets digested How to confront data science I think it will always have an impact There's a lot of talk about which jobs are going away and which jobs are staying but we don't know what will happen in the future so I personally think that there's no need to worry too much about it As AI technology becomes available I think there will be a big difference between those who can and cannot use it well When that time comes when there's a big impact like that, in order to avoid seeing them negatively I think we should take advantage of new things and accept them Well, in two or three years time I believe that it'll be mathematics that will be useful I think it is very necessary to try something new in order to lower the hurdle Since the basis of data science will most likely always be mathematics and statistics to keep up with the various trends of the times I believe maintaining basic math skills are definitely necessary