How AI 'thinks'

By the end of this day you will know how algorithms operate, how to spot them in operation, and how to learn from their logic

AI (1).jpg

Machine learning algorithms


Watch this Friendly Introduction to Machine Learning Algorithms by Luis Serrano, the Machine Learning Nanodegree lead at Udacity.

There are loads of interesting types of algorithms. Try writing down the types that Serrano mentions and match them up with more real world applications. Understanding which types do what will help you pinpoint how you can actually use them.

Teach your own


Now, for a little fun, try out teaching your own machine with Google’s interactive Teachable Machine game.

This game is full of silly gifs and sounds but gives a really good demonstration of how we train neural nets, and as a result, how we ourselves actually learn.

Blind spots 


Watch this TED talk by Blaise Agüera y Arcas, a machine learning scientist at Google, on How computers are learning to be creative.

Algorithms are making the world go by but should be treated with caution. Brainstorm some of the problems which could arise if we can’t see or understand parts of the computer’s thinking process.

Big Data


Watch this TED lesson, Exploration on the Big Data Frontier, by Dr. Tim Smith, software developer at CERN. Follow the activities and take part in the discussion. Also, remember the ‘Quick, Draw’ game you played on Monday? Have a look at the open data sets that make it work.

Data is critical to artificial intelligence: it forms the inputs that AI needs in order to work. So, the quality of the data on which we build algorithms directly affects the quality of the answers they produce. After this activity take some time to think about the data infrastructure in the place where you work. What data is collected, and how securely is it kept?

Lunch break

Spot the algorithm


Spend some time exploring the algorithms that shape your digital experience and thinking about what data they are using.

Try the same keywords on search engines other than Google, e.g. Ecosia, DuckDuckGo or Excite. They all have different data sets. ‘Logic’, ‘freedom’ and ‘piece of pie’ are a few to start off with. Notice the adverts on your Facebook and search engine - do these reflect any recent searches you made? Now try switching off your social media news feed sorting algorithms (instructions here).  Curious about what words you most often type into your phone? Try starting a sentence on a messaging application and then pressing the centre button to autofill the rest.

Finding your algo-rhythm


Take a walk and listen to this podcast by Brian Christian about what we can learn from algorithms about unlocking our own potential. Then read the introduction and fifth chapter of the bestseller Algorithms to Live by that he wrote with Tom Griffiths.

Brian Christian and Tom Griffiths have done some amazing work into what computers and their algorithms can teach us about human thought processes and decision making. His book is a helpful exploration of how we can learn from algorithms in a practical way.



An exercise in algorithmic thinking

Algorithms are a way of linking processes together, like a cooking recipe. Read through this great explanation from Duke University and write down some of your daily tasks as algorithmic instructions to a robot. For example, how might you detail the process of brushing your teeth? Be careful to make no assumptions about the robot’s intuition. It can only do exactly what you tell it to.

Now do the same for more complex tasks in your day, such as writing a report or structuring an email. Does this make you look differently at how you complete these tasks, or even the assumptions you’re making in doing so?

(Still curious?)

If you’d like to find out a bit more about the technical side of how algorithms work, go through this course from Khan Academy. On the social side, look through Brian Christian and Tom Griffith’s book.

This article lists some of the best Youtube videos and TED talks to help you understand more about Big Data. The short clips to help you understand more about the terminology will be particularly useful for navigating this new world of jargon.


Short on time? Want to learn more?

We also run a one-day taught course in London and Oxford if you'd like to meet our experts in person.