From product images to customer information, most businesses today have mountains of data. With the right analysis, this information can tell stakeholders about key trends and potential opportunities to improve their businesses. That said, doing so is easier said than done. Simply put, all that information is too much for humans to analyze on our own. With artificial intelligence (AI), however, we now have the perfect tool to do just that.
New to AI? Download this comprehensive and simplified glossary!
Despite now being one of the most critical aspects of any business, data analytics is still quite difficult. With multiple, often fragmented, data sources, many data analysts have to manually consolidate this data before they can analyze it. Further, one study showed that just preparing the data took up, on average, 60% of an analyst’s time.
This kind of rote task is perfect for AI, with many companies already using AI to moderate and organize their data. In spite of this, in 2017, Gartner found that only 10% of businesses used or planned to use AI for this purpose. In a more recent study with chief information officers (CIOs), data analytics was only just behind AI in a list of technologies they thought would be “game-changers” for their organizations, showing its importance. Interestingly enough, that same study also found that identifying use-cases for AI was the third most cited challenge these CIOs thought their company faced for adopting AI, further indicating a lack of awareness of AI’s value in this area.
So why integrate AI with your data analysis efforts?
1) Volume, volume, volume
As we’ve stated before, one of the most significant advantages of AI is how much data it can process, compared to humans alone. With the emergence of social media, for instance, drug safety teams from pharmaceutical companies must now keep track of and collect relevant drug performance data from both traditional feedback channels, like drug trials, and these more ad hoc untraditional feedback channels. While some companies dealt with this considerable increase in data by outsourcing rote data analysis tasks, many now see AI as a cost-effective solution to this problem. 62% of drug safety experts have already implemented or planned to implement AI in their data analysis process.
Social media has also changed the game in other industries. Recently, a team of researchers at Pulsar, a social-listening platform, sought to analyze the spread of the fashion trends predicted by experts at Vogue Magazine using social media posts made during fashion week. Today, 700+ million photos, shared across social networks, now accompany the 100s of shows at these semi-annual international events.
Pulsar used Clarifai’s computer vision technology, specifically our apparel model, to analyze the content shared in 2018. With AI, they were able to analyze over 1 million Twitter and Instagram posts with the #fashionweek hashtag, and then narrow their focus to posts hashtagged #fashionblogger (over 400,000) and #streetstyle (just under 200,000.) Without AI, the team would have likely missed the majority of such posts, making for a far less comprehensive study.
2) Get a comprehensive view of data
The results of Pulsar’s analysis shows another advantage of utilizing this technology. While AI cannot replace years of experience within an industry, it can give analysts a more complete view of the data at their disposal, allowing them to make more informed decisions.
AI allowed Pulsar to see that not all the trends Vogue experts declared to be relevant for the Spring/Summer 2019 season were actually “trending” among fashion week’s largest audience. While the low adoption rate for “cycle shorts” could be blamed on the weather, other trends like “bows” and “beige” were also missing from the outfits.
They were also able to see how to identify distinct differences between how specific audiences interpreted the trends that were making a dent. Both “fashion bloggers,” i.e., persons who had built careers around their sense of style, and the “street style” audience, which represented what “normal” people wore, incorporated lace (the most adopted of the predicted trends) into their styles.
However, while fashion bloggers often made lace pieces the central component of their outfits (e.g., in cocktail dresses,) “street style” wearers preferred to have it as just one key element, like lace tops.
There are likely many reasons for these discrepancies, after all, which is why the expertise of fashion critics, editors, and buyers remains of the utmost relevance. That said, AI can still offer these experts valuable insights, like the clothing tastes of particular target markets, that they, in turn, can use in their decision-making.
3) You can’t know what you don’t know
Though he is best known for his painting “The Scream,” renowned Norwegian artist Edvard Munch actually completed over 1800 pieces, across various art styles, in his lifetime. Much of his work survives, but only a handful of his portfolio is widely known. As asked by self-proclaimed art nerd Jason Bailey of Artnome, the largest analytical database of known artworks in the world, ”How much can we really know about an artist by seeing only a handful of his works?”
With that in mind, Bailey and his team built a custom model (on top of Clarifai’s General Model), training it on paintings to examine Munch’s full portfolio and learn more about the man behind the art. The results surprised them. For example, while Munch is known for expressing “his tormented inner world” in his work, with AI, the Artnome team was able to dig deep into his extensive, mostly unseen, portfolio and find he was also the creator of many more hopeful art pieces, depicting sunrises and colorful landscapes. Combining visual recognition with their expertise allowed the team to gain a more nuanced view of Munch that contrasted with the reputation he developed from the work he created during more turbulent times in his life.
Interest in AI has ballooned in the last few years, with PwC finding 72% of business leaders identifying AI as being fundamental to their future. Despite this, adoption remains low, with a recent Gartner study finding only 14% of businesses had actually taken steps towards deploying AI. When we consider that many businesses are still unsure of how AI can add to their business, and how challenging it is to bring well-versed AI talent and expertise in-house, it’s easy to see why adoption has been slow. While most companies interested in AI want to use it towards customer engagement and support, these aren’t the only ways AI can help your business. Data analysis is already an area many companies across industries are investing in, and there are few technologies more ready to support these efforts than artificial intelligence.