An AI system which could explain its decisions and outline what the user could change to get a better result would be a prescriptive system and would be a huge step toward finding a common language between human and machine.
From the original article “AI-powered marketing needs interpretability – and collaboration” on The Drum.
This article makes the important point that while AI has great potential to revolutionize the marketing industry, and adoption is high, the application of AI technology on marketing problems hasn’t reached its potential because marketers don’t know what to do with the information -or don’t yet trust the recommendations.
As producers of AI-powered solutions for content marketers this speaks to challenges we face here at MarketChorus and highlights a controversial trend in AI software design.
There are 2 kinds of AI solutions…those that inform and empower human users and those that seek to replace humans with algorithms. We know that software that tells a human expert how to do their job isn’t going to be well received.
But software that serves human operators, saving them time, removing redundancies, and presenting data-driven insights is more useful to a professional audience than Netflix-style automagick recommendations.
Like us, Datasine is building machine learning algorithms to solve problems for marketers, and though they’re focused on a completely set of challenges, there’s a lot discussed in this article that sounds familiar.
[Datasine] achieved this by teaching our AI platform, Connect, to understand human and social terms like “the image looks busy” or “that font is old-fashioned”…by understanding these concepts at [scale]…Connect can understand how each of them drive ad-performance…and communicate this back to the user.
MarketChorus uses natural language processing (NLP) to understand the topics being written about on the web on a daily basis and leverages Twitter data to understand its impact on engaged audiences across the web.
Machine learning gives us the ability to analyze volumes of information that human operators could never manage -but it doesn’t replace the raw intelligence and creativity of real human expertise.
Maybe one day we’ll achieve true AGI (artificial general intelligence) that can have creative input beyond the pattern-matching and predictive analytics machine learning excels at today.
All technology has limitations but humans have been tenacious over-achievers for millennia. AI has only existed for a blink of time in comparison. That’s why we’re committed to the philosophy described in this article.
Like Datasine, we’re building tools that help marketers understand insights rather than just deliver recommendations out of a black box.
