Latest Posts

What’s New in Artificial Intelligence in 2020

Image of motherboard with text "Artificial Intelligence in 2020" on top

The year 2020 has seen a lot of change, and much of that change hasn’t been great. One thing that has changed for the better this year is Artificial Intelligence. AI is continuing to advance rapidly, the way computer-based technology usually does.

How Did We Get Here? 

Artificial Intelligence (AI) became a field of study around 70 years ago. As recently as two years ago, most talks of AI technology from software vendors were considered spurious at best. This suspiciousness is well-deserved. AI has been caught in a vicious Hype Cycle that is only now beginning to approach the phase that Gartner calls the Slope of Enlightenment, where we get to use the good stuff, and all of those promises start to generate real value.

Graph showing the change in expectations over time of new technologies. Created by Gartner.

What’s New in Artificial Intelligence?

AI-Assisted creation is the new jam. Aside from self-driving cars keeping us safe (and making traffic less miserable) and digital assistants giving us instant, easy access to knowledge, 2020 is giving us a glimpse at a new AI capability: AI-assisted creation. 

AI-assisted creation is the ability to create new works with AI assistance. Today, we see the potential of new music, movies, articles, novels, and more being created by AI in our direction.

What if Siri could create new music for you? What if the AI could draw from the foundation of all recorded music and create original works with your direction? After all, music is just math. It will go something like this:

Human: Siri, make a new band. Combine the instrumental styles of The Beatles, The Black Eyed Peas, and Daft Punk, with the lyrical style of Kanye West and Elton John.

AI: I’ve created a new band. Here is the first song.

[Plays song]

Human: Great, now add a bit of Metallica to the vocal style, and change the main vocalist to female.

AI: Here’s the new band.

[Plays revised song]

What We Have Already Seen

This year, this type of AI-assisted creation capability has already been demonstrated by Google’s Deepmind in the form of GPT-3, or Generative Pre-Trained Transformer 3. 

GPT-3 has written articles published in major periodicals, working code in multiple programming languages, and designed web pages and User Interfaces. It’s also been responsible for generating new Machine Learning models, functioning spreadsheets, and more, for users who ask for what they want in plain English. This technology has the potential to disrupt creation itself.

Younger people are ready for this. My 9-year-old daughter uses both Siri and Alexa and asks questions like, “Show me a picture of endangered animals,” or “How long would it take to drive to the moon?” 

It’s amazing to watch her interact with AI and see the knowledge get instantly delivered.

How Artificial Intelligence Assisted Creation Could Change the Future

Some people are concerned that this will limit our ability or tenacity to think and learn for ourselves. I believe the opposite is true. 

In 2020 Artificial Intelligence is reducing the busy work involved in learning. A 4th grader can write a report on an endangered species in an hour with AI, or a week without it. We have to ask ourselves what it means to learn. The manual act of searching, scrolling, reading, and flipping pages, isn’t necessarily where the learning happens. Often it’s just busywork. 

AI is freeing us to learn more and do more in less time. In some sense, my daughter is already using AI-assisted creation to speed up her education. Interestingly, she did this on her own, asking, “Why do I need to write or type if Siri will do it for me?” The appropriate use of this tech could be an incredible advantage that we’ve barely scratched the surface of.

In my next article, I’ll dive into more technical detail on how GPT-3 works, what its limitations are, and what we might see next in GPT-4. That is, unless Deepmind gives me API access, in which case, I’ll be having GPT-3 write all of my future articles.