Concepts of intelligence
By the end of this day you should: be able to explain how scientists are attempting to build an intelligent agent
Watch the first lecture from the HarvardX online course, Computer Science for Business Professionals on ‘Computational Thinking’ (you’ll have to enrol) and take the assignment quiz.
This lecture is a clear, engaging introduction to how the computers you use every day actually work and processes information. Do you think the title computational ‘thinking’ is appropriate?
Now take a walk and ponder how computational ‘thinking’ compares to human thinking.
Aside from the important question you’ll be thinking about, this is also a reminder that you’re much better at both walking and pondering than any robot (yet).
Ask yourself what similarities you can see between coding and natural language (that is, the ways humans speak). Think about how the logic of computer code reflects the natural logic of the way we think and write, such as in narratives and sentence structures.
Investigate how code shapes even the websites you browse by right-clicking on any site you like and selecting ‘view source’ or ‘page source’ from the drop down menu.
Machine language is everywhere under the surface of what we see, so knowing some code can be really helpful especially if you run a website or have to collect customer data.
Read chapter two of Nick Bostrom’s book Superintelligence.
This chapter talks about the various ways scientists are looking to integrate human and machine thinking. Bostrom runs the Future of Humanity Institute at the University of Oxford, a department dedicated to investigating where society is heading and what problems it may face. Basically it’s his job to worry, so bear that in mind as you read.
It’s important to understand how we are teaching computers to think based on our own brains because the process of constructing these systems artificially opens up the question of what we actually know, and still have to learn, about our own intelligence.
Read more about deep learning with this article from Machine Learning Mastery, which gives a thorough overview of the topic with quotes from world experts. Now read through this Github site that explains some more about neural networks in particular and play with their simulation, which lets you train your own. You can also see a visualisation of how neural networks act with this handwriting game.
Deep learning is a subfield of machine learning partly inspired by the study of the the biological nervous system. The artificial neural networks that data scientists are developing are one of the clearest examples of how we are modelling computers on our own brains to create artificial intelligence. These articles and games explain how scientists are exploring the area, and demonstrate how the technology works.
What could a machine do for you? Process improvement.
Firstly, read this article from Wired about Elon Musk’s Neuralink project and think about what questions this raises. What boundaries we should be drawing, if any? Also think about the word ‘intelligence’ - what do we really mean by that word? And are our definitions challenged by differences between natural and artificial intelligence?
Given what you’ve learned today about computational thinking, think about whether there are tasks in your life or work that could be delegated to an AI system (e.g. paperwork, taxes, data searching)? Be thinking about what it now means to work intelligently.
If you’d like to learn from more about coding there are several good MOOCs from edX (e.g. CS50), Coursera, or Udacity in addition to the Code Academy site that you’ve seen today. They all have different styles, so take a look to see which one best suits your learning style and timeframe.
To find out more about neural networks and deep learning, this online book by Michael Nielsen goes into more depth about the core concepts and how the technology is poised to provide solutions to many problems in image recognition, speech recognition, and natural language processing.