There are two common debates you’ll hear over Artificial Intelligence (AI) right now.
The first is the “end of days” argument we’ve covered here a few times. You know: robots taking over the world, annihilating mankind, taking all our jobs, or generally just making us feel really uncomfortable.
On the other side of the spectrum, a different argument rages over whether AI can even be made to think independently or creatively in the first place. Or at least, creatively enough to do things like write a novel, or produce more than very basic press releases for news wire services. So far, the answer is mostly no, with some limited exceptions — although there’s no telling what the future will bring.
One graphic — and frankly, absolutely hilarious — illustration of how difficult it is to actually get AI to think creatively is this short film below. The movie’s script was created by an AI “neural network,” and is acted out precisely as the machine (which named itself “Benjamin”, by the way) intended.
The most incredible thing about this is how the actors all managed to keep a straight face throughout.
The second most incredible thing is that they successfully memorized the… er… script.
Beyond the obvious comedy value, this is a great example of two important qualifications to all the hype about AI.
First: AI technology is still relatively young. AI can help us do some truly amazing things, but it’s not even close to being as genuinely intelligent or creative as human beings — let alone superior to us.
But what this short film more clearly demonstrates is how reliant AI is on data. Regardless of how complex the technology is, if you don’t feed it enough of the right data, the results will be of little to no use.
With AI, Data Quality and Quantity Both Matter
How does a computer independently create a screenplay for a sci-fi movie, no matter how full of gibberish it is? By watching — or rather, by being fed — a large sample of movies from the same genre. For “Benjamin”, that included dozens of blockbusters including Highlander Endgame, Ghostbusters, Interstellar, War of the Worlds and The Fifth Element.
In this case, it’s possible the sample wasn’t big enough to “teach” the AI model to create a coherent screenplay — although it’s worth noting that the machine did create at least some semblance of science fiction.
In fact, the amount of data needed to teach a computer to independently write like a human is eye-wateringly enormous. That’s what one journalist found out when conducting a similar AI experiment at around the same time as Benjamin’s creators — in her case, to find out whether AI could write a long-form news article.
But more likely, it’s possible that when it comes to general or creative intelligence, the right kind of data might not even exist in the first place. Specifically: emotions that are integral to storytelling — like fear, love, hate, excitement, jealousy, anger, etc — can’t just be fed into an AI algorithm. When they are, the results are hilarious precisely because it’s clear the machine isn’t able to process these kinds of feelings without actually experiencing them — something AI can’t do (at least at the moment).
To quote Jaime Carbonell, the director of the Language Technologies Institute at Carnegie Mellon University:
“What the machines can do is express weather forecasts based on meteorological data, or a baseball narrative based on the scoresheets—pretty good writing that is almost indistinguishable from human writing when reporting facts in a formulaic style.
“But the question that doesn’t have an answer… [Is] can a machine acquire general intelligence? So far, the answer has been, ‘no.’ There are things like writing fiction or writing poetry where it’s not clear what it means for a computer to be able to do it, since the machine can not directly experience the emotions you’re trying to convey. It’s difficult to fathom how you could generate genuine creative writing in that sense. There are categories of activities in which it doesn’t really make sense to train the computer to do it. You would just get some ersatz version of a human.”
The Right Marketing Data is Out There
Fortunately for marketers and countless other professionals, in many cases the right kind of data does indeed exist. The trick is getting hold of it, while avoiding faulty data that could skew the results.
That’s the idea behind Leadspace’s Audience Management Platform: namely, that neither predictive analytics nor “big data” alone can provide real, long-term value. Yet when the two are combined, the result can provide enormous value to businesses, and solve many pressing real-world problems.
For example, similar Deep Learning “neural” AI can be used to learn the anatomy of a company’s ideal customers, and then provide lists of new prospects who mirror or closely resemble those customers.
That Look-alike Modeling solution is ideally suited for marketing methods like account-based marketing (ABM), which require highly strategic, targeted lists of accounts to go after.
For an example of how it works, read the free case study below:
Picture credit: Pixabay | CC0 Creative Commons
The post What This Hilarious AI Experiment Can Teach B2B Marketers About Their MarTech appeared first on Leadspace.