Earlier this year, @DeepDrumpf joined Twitter, and if you didn’t look closely enough you may have confused its tweets with @realDonaldTrump‘s actual Twitter gems.
[@POTUS has @Spotify? I have sex tape playlists.] They call it amazing and vital for USA. Soon we’ll have no party to change.
@HillaryClinton [Hillary Clinton] was all talk. I was screaming — jobs and extremists, not policy. But I won. @ChadHGriffin @HillaryClinton #debates2016
That’s because @DeepDrumpf is an AI-based neural network that is trained on Donald Trump’s transcripts. What does that mean?
Well, if you’ve tried apps like Digit or caught stories like this, you may know that it’s now possible to mimic human speech based on algorithms and probabilities. So how close are we to Her-like moments where we communicate in full conversations with robots?
On the latest SeedStories podcast, we decided to sit down with @DeepDrumpf‘s creator Brad Hayes to find out. Brad is a Roboticist, an AI Researcher and Postdoctoral Associate within MIT’s Interactive Robotics Group.
In our discussion, we touch on what this all means in layman’s terms, how he developed @DeepDrumf, what’s next for artificial intelligence and robotics, and if we need to be worried about a doomsday scenario where AI takes over.
Specifically, we hit on:
- How he made @DeepDrumf and the changes he needed to make to humanize the bot
- How this compares to the Facebook algorithms that control the news on its platform
- What commercial uses he sees for these types of neural networks in the future
- If it is dangerous tying our emotions to mathematically created conversing
- How far away are we from a robot actually talking and carrying on meaningful conversations as we saw in the movie Her
- What he’s most excited for in AI and robotics, and the new trends in the space that the general public isn’t even aware of yet
- His thoughts on AI-driven doomsday scenarios
Enjoy!
Podcast: Play in new window | Download
Subscribe: Apple Podcasts | RSS
