AI : What role in tomorrow Business ? AI Series #2

This is the second piece of our AI Series. In this piece, we’ll go through different AI applications across several industries.

If you’re still unfamiliar with AI terms such as Neural Networks, Deep Learning or unsupervised learning, I suggest you starting by reading our first piece. Even though a large part of the press and AI expert are using AI to describe any data science including predictive analysis, here we are referring to more advanced capabilities.

Machine learning capabilities have an almost infinite numbers of use cases. Neural networks can potentially solve any discrete problem; as long as there is a lot of data and the answer needs to be the better prediction of outcomes. Deep learning has changed the way predictive analysis are made.

Deep learning is what has made possible natural communication, personal assistants, and precise image & video recognition.

Natural Language Processing

One of the most common application of AI is what is called Natural Language Processing (NLP). Roughly speaking, NLP technology is gathering communication data, either text, and some of the most advanced one voice and video. Once gathered, this data is analyzed and models are created to make sense of this data. Software are then understanding what is said by the text/video and able to use it for different purposes.

Image and voice recognition are amazing training data for deep learning softwares. Deep learning can be applied on both structured and unstructured data. Structured data is when one knows already the type of data the software will get.

Neural AI techniques excel at analyzing image, video, and audio data types because of their complex, multi-dimensional nature, known by experts as “high dimensionality.”

For example, when applied on online reviews written by consumers, this technique will gather & analyze different type of data: The review will be text, and then it will get all the user data (gender, location etc.).

AI is then used to understand what the words of the text means, and what are the user sentiment while using these words.This type of learning is supervised, meaning a team of human is helping the AI understands the link between phrases and feelings.

Revuze, graduated from the Nielsen Innovate incubator in Caesarea, Israel has turned its attention to product experience management (PEM), enabling clients to “measure customer perception of the holistic product and service experience.”

Revuze analyses aren’t limited to reviews; the company’s tech applies NLP for topic, keyword, and sentiment extraction from any form of textual interactions, including survey responses, call-center text, and social media. Generated via semi-supervised machine learning, source text is gathered, sorted and analyzed to provide brand with granular information about products and user experience.

Retail: a natural market for AI applications

As revuze example points out, one of the main market benefiting of AI today is retail. According to McKinsey, AI applications in retail represent a potential market of  $0.8T, followed by CPG, a $486Bn market.

Indeed, retailers are creating omni-channel personalized experience using Machine-learning technology. Even if we will dig deeper on this market applications in our next articles, we can’t bypass it here. So allow me to introduce some basic use of AI for online merchants.

Marketers use NLP to gather data on customers, build target audiences, and optimise the omnichannel experience. The more data companies gather, the easier it will be to apply neural networks to any of the challenges they have along the entire journey to conversion and customer lifecycle.

Dynamic Yield is applying machine learning algorithms to build model that understands consumer preferences and create micro-segment of user having similar behavior. When I started working on Marketing, we built segment depending on location, sometimes on time-since-last-purchased, or AOV.

Now, thanks to companies like Dynamic Yield, marketers are able to build very specific micro-segment, encapsulating data from different sources, both internal and external. Then, this micro-segment will be assigned to different consumer journeys.

This is by using similar technology than your Amazon feed doesn’t look like mine. At least it shouldn’t. Personal recommendations, tailored banners, dynamic menu are becoming the norms for online retailers. People are used to be offered content that fits their need and not look for anything more than a few seconds.

Our cognitive biases lead us to focus 70% of our brain looking for the negative aspects, so when we are online, the minor bug can push us out. For example, payment processing inefficiencies are one of the largest cause of cart drop. And a website visitor with a bad experience is more difficult to convince a second time.

This is why online merchants need to pay attention to every detail, opening opportunities for large set of specialised services resolving one aspect of the customer journey at a time.

AI application in fintech

In fintech, we have found also interesting use cases for fraud-detection, but it’s the topic of another article. Riskified, one of our Tel aviv neighbours, are bumping conversion rate by making the payment experience safer seamlessly.

Described as an AI approval engine, the company is helping merchants avoid false positive and let legitimate consumers process their payment.

Insurance companies & banks also start to use AI as prediction tool for user behavior. They can now sort user according to more accurate predictions, and hopefully create the fairest offers for them.

AI applications across industries

Marketing & Sales services are one of the largest market today, accounting for $1.4T revenues last year. Companies are using AI technology and NLP to identify data structure, understand pattern of behavior, create predictive analysis, leading to recommendations, and sometimes engaging actions.

According to Keyrus Innovation Factory examination of AI application in businesses, one of the highest added value for companies is today besides Marketing & Sales is in operational functions such as supply-chain management & manufacturing.

Of course what we’ll hear about are Virtual Assistant, like Google Assistant that is able to make an appointment for you and have a conversation with a human-being. And this is certainly impressive. After all, the technology needs to master speech recognition & dialogue.

AI will open the doors for unthinkable opportunities

But these giants are using virtual assistant as a nice tool for consumers to have, in order for them to gather more data, and thus making their back-end AI maximise their revenues. But scepticism and doubts about AI are the topic of another article.

We’ve already discussed Automotive application and autonomous vehicle in a previous article, but it is certainly one of the best known example of machine learning. The amazing part, sometimes unknown, is that a large number of data is still processed using supervised learning.

It means AV companies are hiring very large team of people actually processing data manually and giving answers to the software to help him refine its prediction & decision models.

Analytical techniques are accounting for over ten trillion dollars annually, and AI powered intelligence across 19 industries could potentially represent 40%, or $3.5 trillion. In addition to the markets mentioned above, existing applications are already creating new standard in Logistics, Healthcare, Cyber, Agritech & Climate study and of course Aerospace & drones.  

Sooner or later, every company will need to examine its mix of functions and find the most pertinent and attractive opportunities to use AI. Over the next few year, an explosion of use cases will rises where companies will experiment different ways to use AI, exploring things they could never do on their own.

The exciting developments in AI delivering jumps in the accuracy of classification and prediction. As consumer data & intelligence become more and more available, deep learning will become more and more efficient, fast and accurate. We will then see plethora of use cases and firms in every industry will need to collaborate with innovative companies.

To go further, here are some sources that inspired me:

• https://www.nytimes.com/2017/07/29/opinion/sunday/artificial-intelligence-is-stuck-heres-how-to-move-it-forward.html?_r=0
• https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/what-ai-can-and-cant-do-yet-for-your-business
• https://www.mckinsey.com/featured-insights/artificial-intelligence/visualizing-the-uses-and-potential-impact-of-ai-and-other-analytics
• https://martechseries.com/mts-insights/interviews/interview-jeremy-fain-ceo-co-founder-cognitiv/
• https://www.thomsonreuters.com/en/reports/2018-ai-predictions.html