VIDEO
HIGHLIGHT
Log InSign up
The Future (and Past) of AI - Kathryn Hume (integrate.ai)
SummaryTranscriptChat

The Future (and Past) of AI - Kathryn Hume (integrate.ai)

Introduction

The speaker introduces the topic of AI and explains why predicting the future is difficult.

  • Humans are bad at predicting the future because we tend to imagine it from our present perspective.
  • Language used to describe technology reflects our historical imagination.
  • The speaker will take a different approach by walking through the fictional life of someone born in Montreal today and how AI technologies will impact their life.

Historical Imagination

The speaker discusses how language used to describe technology reflects our historical imagination.

  • Language used to describe technology reflects our historical imagination.
  • Example: "horsepower" was originally used to describe engines in cars because adding more horses to a horse-drawn cart was seen as a way of going faster.
  • Our bad historical imagination is reflected in the words we use to describe technology.

Fictional Life of John Sebastian

The speaker introduces John Sebastian, a fictional character whose life will be used as an example of how AI technologies will impact people's lives.

  • John Sebastian is a fictional character who was born in Montreal on July 24th, 2017.
  • His life will be used as an example of how AI technologies will impact people's lives.

Using Natural Language Generation for Medical Data

The speaker discusses how natural language generation can be used to represent medical data in clear language that is relative to the cares and concerns of an individual.

  • When John Sebastian was born, he had some health issues and was hooked up to an echocardiogram machine.
  • Thanks to artificial intelligence technologies, data from the machine can now be represented in more friendly human language using natural language generation (NLG).
  • NLG can personalize communication for different users based on their needs and concerns.

Image Recognition for Learning

The speaker discusses how image recognition technologies can be used for learning.

  • When John Sebastian was in preschool, he went on field trips to learn about the natural world.
  • Thanks to new image recognition technologies, he could learn about flowers and other things more easily.
  • Image recognition technologies can help people learn by identifying objects and providing information about them.

Augmented Reality Tools

The transcript discusses how augmented reality tools are being built using image recognition techniques to help people learn about the natural world and other subjects.

Learning with Augmented Reality

  • Augmented reality tools are being developed to help people learn about the natural world and other subjects.
  • The Louvre works with cultural institutions and technology companies like Google to create new shows that use style transfer applications powered by AI to recreate paintings in different styles.
  • Jean-Pierre, a young boy, becomes curious about whether there is any domain of human activity that cannot be replicated by AI after seeing how easily machines can replicate the Mona Lisa. He turns to music as a potential area where AI may not be able to capture its complexity.
  • Jean-Pierre's second-grade music teacher introduces him to AI techniques that use vast corpuses of past music to discern patterns over long time sequences and create more sophisticated music.

Music and Artificial Intelligence

This section explores Jean-Pierre's journey into learning music through artificial intelligence.

Learning Music with AI

  • Jean-Pierre learns some simple rules of melody but finds that music is actually quite complex due to motifs started at the beginning of a piece that get carried on later on and references to past work.
  • His second-grade music teacher introduces him to AI techniques that take vast corpuses of past music, set them through an engine, discern out their systems, and create more sophisticated music.
  • As a seven-year-old, Jean Sebastian engages in a dialogue with a computer where he can play a sequence of notes from Mozart sonatas or other pieces, and the computer can replicate those notes and play along with him.

Self-Driving Cars

This section discusses the impact of self-driving cars on society and how they are being regulated.

The Impact of Self-Driving Cars

  • Jean-Pierre never has to have his parents carry him around because he's able to be taken in a self-driving car. However, his parents had to pay a high price for liability insurance associated with the self-driving car that they got.
  • Regulators and policymakers are still fighting about the potential danger attributed to society by bringing on these new technologies.
  • There are debates about who should be found at fault if an accident occurs due to human error or AI error in self-driving cars.

Thinking about AI from a different perspective

In this section, the speaker talks about how we should think of computers as being smarter than humans instead of just gradually progressing upon our capabilities. He also shares some personal anecdotes to illustrate his point.

Jashub's awkward phase

  • Jashub enters into his awkward phase in his teens.
  • The speaker shares a personal anecdote about having a similar experience when he was in second grade.

Online dating in the future

  • In the future, online dating will allow people to walk by someone in real life and take a picture to see if they're single.
  • There will be a confidence rate on the extent to which said person is compatible with us.
  • Virtual reality dates will become more common.

History of AI algorithms

In this section, the speaker talks about the history of AI algorithms and how they have progressed over time.

Learning from past examples

  • The algorithms that are powering the AI revolution today all date back to the 1950s.
  • Claude Shannon built an amazing system that trained a mouse to go through a maze using machine learning algorithm in 1952.
  • Arthur Samuel built the first machine learning system to beat the game of checkers in 1956.
  • Frank Rosenblatt was inspired by the architecture of the human mind to make what he called a mark 1 perceptron which was hardware with wires connecting together nodes of computation and these wires could classify images according to what objects were in them.
  • Geoffrey Hinton realized that we could take these once Hardware systems Hardware perceptrons and mimicked them in software.
Video description

Startupfest 2017 The history of AI in the 20th century was like the weather in Montreal: beautiful, hopeful summers followed were followed by long, depressing winters. In the 1950s, a series of research breakthroughs made the technology community believe smart machines would soon transform human work and lives. Nearly 60 years later, we're singing the same tune, just to a slightly more electronic beat. Will things be different this time or will the AI hype cycle quickly crash? This talk will explore the near future of AI, explaining why the field is exciting, how to succeed (and avoid failure) applying AI in a startup, and what commercial opportunities lie ahead. Startupfest is Canada's original startup event, bringing over 7000 founders, investors, accelerators, and community leaders from across the globe to Montreal, Canada every July. The event is one week of unbeatable networking, education, and parties in an iconic festival format. Learn more at startupfest.com Must follow Startupfest links: - Official YouTube Page: https://www.youtube.com/user/Startupfest - Facebook: https://www.facebook.com/internationalstartupfestival/ - Twitter: https://twitter.com/startupfest - Instagram: https://www.instagram.com/startupfest/ - Linkedin Group: https://ca.linkedin.com/company/startupfest - SlideShare: https://de.slideshare.net/startupfest