Humans- Non-Artificial Un-Intelligence
Blank 101’s maiden voyage into the world of technical talks witnessed an NLH room full of enthusiastic tech aficionados on the 22nd of August. The hour-long talk on Machine Learning and AI started with a brief introduction to the Turing test for intelligence in a computer. This test which checks whether or not an unbiased judge can distinguish the computer’s replies to certain questions from a human’s. Siri would probably flunk this.
The speakers talked about how Machine Learning and AI are directly linked to human psychology, referring to the Theory of Mind. Another topic that was touched upon was self-awareness in machines, computers, and automatons, which would be an extension of the Theory of Mind. Along the same train of thought, the next speaker elaborated upon how AI learns to perform different functions the same way humans do- by experience. Google’s AI DeepMind figured out how to play the infamously difficult Atari game- Montezuma’s Revenge by becoming “curious” enough to want to play the game. The team of researchers programmed artificial curiosity into the AI by giving it rewards for exploring more of the game’s virtual world.
The seminar then took to many real-life examples of Artificial Intelligence in places we’d never expect it to be. This included Google AdSense and Cambridge Analytica- which provides psychographic profiling as a method of targeting an audience and selling a product- be it cereal, or the then President-elect of the United States of America.
The final stage of the talk touched upon the topic of AI going rogue, scattered with examples like TayTweets- Microsoft’s AI twitter chatbot described as an experiment in “conversational understanding.” The more you chat with Tay, said Microsoft, the smarter it gets, learning to engage people through “casual and playful conversation.”
Unfortunately, it took only 24 hours for the bot to get corrupted and start spewing racist and misogynistic statements because the internet is not a very nice place. In conclusion, the speaker explained how AI would never as much ‘turn evil’ as it would ‘turn competent with goals misaligned with ours’.
Present day technology may take years to reach even an iota of the complexity displayed by fictional supercomputers like Deep Thought from the Hitchhiker’s Guide to the Galaxy. Regardless, the team at Blank 101 ensured that every person left the room with their mind racing with ideas to devise artificial intelligence that could one day give us the answer to Life, the Universe, and Everything- preferably in non-integer format.