Why I’m Volunteering to Train an AI

Artificial Intelligence is looming. And I don't mean in a "100 years from now we're all screwed" kind of way. More of a "batten down the hatches, it's coming sooner than you think" kind of way.

In recent months AI has come to the forefront of discussions around where tech is heading – and how impactful it will be on nearly every facet of our lives. But I think the point that most people are unaware of is just how imminently the effects will be felt.

Artificial Intelligence

In a recent survey of the Artificial Intelligence community1 , the predictions of when we should expect to see the arrival of an Artificial Super Intelligence2 are staggering:

  • Median optimistic year (10% likelihood): 2022
  • Median realistic year (50% likelihood): 2040
  • Median pessimistic year (90% likelihood): 2075

    Meaning, realistically we should expect the arrival of an intelligence beyond the comprehension of our puny biological brains by 2040. Or, if you’re on the optimistic end of the spectrum, in just 6 years.

    To sum up the extremely comprehensive (and brilliant) two-part from Wait But Why’s Tim Urban: the key factors in this race are who gets there first, if we’ll even know about its development, and how the AI will behave in regards to us humans.

    It goes without saying that the major tech kingpins have thrown their hats in the ring to "get there" first and push past the current day Artificial Narrow Intelligence. Facebook, Google, Microsoft, Apple – they're all devoting enormous resources to development in this area.

    But what's really interesting is the emergence of small scale startups who are just as bleeding edge as the big guys.


    Recently an opportunity with a Norwegian startup called Iris AI caught my eye. The startup was looking for AI Trainers to work with Iris, their budding AI. The small team is shooting big, aiming to build an artificial intelligence which not only has access to the entirety of mankind’s scientific research, but is able to then carry out experiments independently.

    Think about that.

    An intelligence that can test any hypothesis you throw at it and provide you with the findings based on the collective scientific knowledge of all published research.

    “We want to democratize access to scientific knowledge. The first step is a science assistant leveraging AI and the crowd to help users map out and find relevant scientific knowledge”

    Maria Ritola, Iris AI co-founder

    In these early days, Iris is busy consuming TED talks; analysing the transcripts of talks, and finding the peer reviewed research backing up the content of the talk. The role of the Trainers is to “course correct” – helping adjust the key topics Iris reads into each talk, or suggesting new ones that it may have missed.

    The ramifications of the Iris AI team making their lofty ideas a reality are just impossible to wrap your head around.3

    But exactly why I'm spending an hour a week training Iris is just my excitable curiosity. How will "small" AI projects like this one develop in comparison to the giants? What does the "2.0" of a project like this look like, and how big are the leaps between the steps?

    Personally, I’m optimistic about what the future holds for upstart AI projects like Iris. I love to see any niche idea go big, challenging the status quo.

    How the world will react when the first Artificial Super Intelligence arrives is super interesting. Will we run for the woods or hang on to the life we know and hope for the best?

    Hopefully Iris will remember me as one of the humans who helped nudge it in the right direction and play nice.4

    Must Read

    Wait But Why, The AI Revolution

  • Part 1: The Road to Superintelligence
  • Part 2: Our Immortality or Extinction

    World Economic Forum

  • No, the robots are not about to rise up and destroy us all
  • 5 ways artificial intelligence will disrupt science
  • Have we hit a major artificial intelligence milestone?

    1. Future Progress in Artificial Intelligence: A Survey of Expert Opinion by Vincent C. Müller Nick and Bostrom

    2. Using the defintion of Nick Bostrom, of Oxford's Future of Humanity Institute, in his paper How Long Before Superintelligence:

      By a "superintelligence" we mean an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills. This definition leaves open how the superintelligence is implemented: it could be a digital computer, an ensemble of networked computers, cultured cortical tissue or what have you. It also leaves open whether the superintelligence is conscious and has subjective experiences.

    3. In fact, the World Economnic Forum recently highlighted Iris AI in it's recent article 5 ways artificial intelligence will disrupt science

    4. Speaking of getting on the AI's good side...Iris AI is taking on new AI Trainers! If you're interested, register here.