09 Mar 2017
The fascinating history of AI
The history of AI, and the genesis of our eventual robot overlords, began on a cold day at Dartmouth College in 1956.
Of course, “robot overlords” were only the worst-case scenario. At the time, computers were nowhere near as powerful as they are today, and as this blog post is being written, humans are still in control of society.
However, the Dartmouth conference was where the term “AI”, aka “artificial intelligence”, was first used. The conference’s main idea was that if you could describe learning and intelligence in precise detail, you could create AI
It’s uncertain how much impact this specific hypothesis had on the field and history of AI, but the conference itself was instrumental in legitimising AI as a field of study.
For decades after the conference, AI was to be the subject of mostly theoretical research, funded by a United States government eager to reap the benefits of computerised super intelligence.
AI researchers throughout the ’60s, ’70s and ’80s would generally follow a standard pattern still recognisable today:
- Boldly (and unrealistically) claim that supercomputers with AI would be possible within 10 years
- Get massive amounts of funding from the U.S. government
- Achieve no tangible results
- Get funding cut
This pattern repeated several times over the history of AI, but no real results were had because the computers of the day simply weren’t powerful enough to do much with A.I…. until the late 90s, that is.
Our chess loving AI buddies
It’s 1996. Famous chess champion Garry Kasparov is staring so intently at a chess board that he’s practically burning a hole in the wood.
“Checkmate,” Garry says as he moves the final piece.
Kasparov was facing IBM’s Deep Blue, an AI that played chess through the use of massive processing power. Deep Blue used a “brute force” method for computing possible chess moves -- a skill that won 2 matches against the chess Grandmaster.
However, it wasn’t enough. Kasparov won 4 matches to Deep Blue’s 2, sealing away humanity’s title as the chess champions on Earth.
That is, until the rematch the following year. Deep Blue was upgraded, and had a grudge burning deep in its robotic soul. After winning the first match of the series, Kasparov laid a trap for the computer that had worked a year earlier.
However, Deep Blue was now smarter and had become accustomed to Garry’s tricks -- the AI didn’t’ fall for the bait. Apparently Gary was so shaken by this that he lost all 5 matches after that point.
But chess is such a constrained game. Computers still weren’t REALLY intelligent -- after all, they couldn’t even understand human language… right?
Watson ruins game shows for everyone
Fast forward to 2011. IBM developed yet another AI named “Watson”, which was competing against two Jeopardy champions who had racked up $5 million dollars in combined winnings over their trivia careers.
At the beginning of the match, it was a fair fight -- Watson was smart, but he had trouble with Harry Potter and with questions related to decades (apparently magic and AI don’t mix).
However, as the game went on, Watson pulled out way in front.
By the end of the match, the final score wasn’t even close:
Watson -- $77,147
Rutter -- $21,600
Jennings -- $24,000
Ken Jennings, who had won 74 Jeopardy games in a row before meeting his cybernetic downfall, wrote on his Final Jeopardy answer, “I, for one, welcome our new computer overlords.”
Jennings shouldn’t feel too bad though. Apparently it took 20 researchers three years to create the software Watson used to win at Jeopardy.
Microsoft’s racist chatbot
However, the world of AI isn’t all fun and games.
In March of 2016, Microsoft launched a Twitter chat bot called “Tay” that was supposed to engage 18-24 year-olds. “Tay is designed to engage and entertain people where they connect with each other online through casual and playful conversation,” Microsoft said. “The more you chat with Tay the smarter she gets.”
What Microsoft didn’t take into account was just how far trolls would go in an effort to mess with Tay.
As The Verge reports, “Pretty soon after Tay launched, people starting tweeting the bot with all sorts of misogynistic, racist, and Trumpist remarks.
And Tay — being essentially a robot parrot with an internet connection — started repeating these sentiments back to users, proving correct that old programming adage: flaming garbage pile in, flaming garbage pile out.”
Modern AI and marketing
Standing a few months into 2017, what does the history of AI portend for the future? Specifically, the future of marketing (we’ll leave the doomsday scenarios to Hollywood for now).
At Inbound 2016, the co-founders of HubSpot predicted that:
- Chatbots would soon replace many of the functions of a consumer facing website. Instead of consumers searching through your website for useful content, they’ll tell a chatbot what their issue is for immediate direction.
For example, HubSpot has built a chatbot called “GrowthBot” that can answer a variety of marketing and lead generation related questions.
- AI is going to replace a lot of the boring, routine work that current marketers have to deal with. Soon you’ll be able to say goodbye to tasks like fixing broken links, or updating your CRM manually.
- Voice input will start to be the primary way people interact with computers. As voice recognition AI gets better and better, people will start to speak to their computers instead of typing on them. This is great -- speaking is much more natural, and doesn’t lead to carpal tunnel.
It’s unclear exactly how else AI will change our lives, as we’re very early in the development of truly useful, ubiquitous AI
However, two things seem certain:
- The history of AI is still being written
- We in the marketing field will soon be increasingly interdependent on our AI friends.