Having artificial intelligence powering your brand’s social analytics is table stakes, but AI comes in a variety of iterations these days – and yes, there’s such a thing as “old AI” at this point! And when smart AI isn’t all it’s cracked up to be, your brand’s health is at risk.
So, here’s a handy way to think about it, using some popular AI failures as a framework. It will certainly provide some talking points for your next discussion with your vendor.
What does a social analytics’ AI capabilities have to do with these fictional robots though?
There are flaws in each AI shown – deceptively bad characteristics underlying each, rather. And these flaws line up with things we’ve heard from and about some “smart” AI-powered analytics tools on the market. This is just a quick, relatable way to illustrate each. A way to help brands nail down what matters when comparing tools.
AI That’s Hyper-focused on Few Sources
David of A.I. is “imprinted” to his human mom and hyper-focuses on her in an unhealthy way. This ultimately leads to an equal extreme – David being abandoned.
There are social analytics AI options that share this fatal flaw as it relates to insight. They are apparently “imprinted” to focus on a short list of sources and/or hold certain insight sources above others in an unhealthy (for your brand) way. This provides a super skewed perspective. And one that leaves out entire segments.
This AI’s inability to accommodate other sources is a leading reason for user abandonment. And it typically happens as the holiday season gets into full swing (so we’ve been gearing up!).
This hyper-focus on too few sources results in brands relying on limited insight. And that leads to correspondingly limited campaign strategy.
Combat this by asking very specifically about their data sources. If they’re not capturing and analyzing insight from a wide variety of social web sources, run as if a creepy little AI kid is chasing you.
Want the gold standard to compare it against? Add us to that vendor list and start/end your search here!
Missing the Mark by a Ridiculous Amount
Kryten from Red Dwarf made odd miscalculations pretty consistently in its quest to be independent. Probably its most startling error was when the selfless robot was found caring for dead humans he misunderstood to be injured, but alive. Maybe miscalculation is too kind – but it’s certainly fitting, as it was based on misplaced logic. An AI that thought itself more advanced than it actually was.
And this is where we find many “smart” AI options today. So close, but missing crucial components to connect the dots. And certainly not next generation tools able to automatically discover themes without user bias mucking it up.
They’re a lot like our other humanistic AI failure – the Replicants from Blade Runner. These confused beings never realize their AI’s full potential much like confused social analytics tools.
Rather, these tools offer a confused combination of powerful ingredients. And the resulting mishmash is harmful to brands requiring confident output (which should be every brand). Also, like Replicants, these AI “skin jobs” can be found out with a few specific questions, like:
- Explain your AI to me – is it using best in class machine learning, deep learning and expert systems for intelligent processing and analysis of social media and unstructured data at scale?
- How fast is it compared to other tools?
- What is the history of your company’s AI capabilities? How has it progressed from your company’s inception to today?
For your reference: Our next generation artificial intelligence is optimized to analyze the full corpus of all data sources without sampling. Complex queries are our specialty, and our speed is unmatched.
But it isn’t all about processing speed, of course. Smart AI is valuable if you’re willing to wait a little bit for it. Sometimes it isn’t smart, is the thing.
Smart AI That . . . Isn’t
Online changes – fast. If you don’t stop and look around once in a while, you’ll miss it. And if you’re using the wrong tool to track it, you definitely will! And sending the wrong tool to do the job will result in failure. Yes, like the Terminator.
The Terminator T800 was supposed to be so advanced, so amazing – but it relied on the phone book to find Sarah Connor. And that’s a lot like a social analytics tool that used to require 15 instances to train its AI and now needs 30 instances to train it. So, it’s actually devolving. At least the Terminator gets smarter with each iteration!
To be fair to Arnold’s Terminator though, dumb AI is probably more akin to the ED-209 from Robocop. A big, clunky bit of business that’s skilled at one specific thing – killing, but can’t sort out how to walk down the stairs.
And that’s the scary part about these tools (and the similarity to a clumsy, killing robot) – relying on it to have a nuanced understanding of what lies beneath surface insight is dangerous. And the building blocks of consumer experience understanding – that nuanced understanding – is essential to capturing consumers’ true motivations.
Understanding Sentiment Specifics
The sentiment drivers that inspire specific behaviors are entirely relevant to brands’ efforts and those drivers have to be informed by powerful, accurate and transparent next generation artificial intelligence.
Without it, brands have a false sense of understanding based on that one super skill the AI demonstrates. And it will inevitably take your brand for a hard tumble down the metaphorical stairs.
Some AI chugs along, barely managing to pull through though – and that’s probably the one your need to watch for closest of all. Because when push comes to shove, they stall out and it usually happens at the worst times, from a sales standpoint. Times when their system is under stress, like the holidays.
Smart AI That Struggles with API Updates
What were you doing last holiday season? If you were using a social analytics tool that wasn’t NetBase, you may have been frustratingly unable to access Instagram insight for a good chunk of time.
These tools are like the Cylons in Battlestar Galactica, who created beings that eventually destroyed them. These tools aren’t powerful, but inspire disruptors in the space that are – and that inevitably lead to their destruction.
With lots of competitors popping up in the analytics space, and lip service around what tools can do, there are some serious contenders. Tools that aren’t able to live up to the hype they create are increasingly called out and that’s a good thing. We welcome exploration of our capabilities.
Because, if our AI resembled any of these, it would be Marvin from The Hitchhiker’s Guide to the Galaxy. It’s super smart and increasingly aware – but certainly not depressed from underuse, as we keep it challenged with constant upgrades. And our team of savvy engineers put it through its paces nonstop.
That is something we’re always happy to show off. Want an in-depth tour of its capabilities? Book a demo today!