The real value for your customers and business
Is your organization using Artificial Intelligence (AI) already? How does it work? What should you know? AI expert Mieke De Ketelaere advises organizations that use AI to choose a multidisciplinary approach.
All about AI:
This article has been written based on a live presentation. Visit “All eyes on B2B” to enjoy the recording and other inspirational videos.
What is Artificial Intelligence?
We will go back in time to understand Artificial Intelligence (AI) completely. AI started in the 1950s as an academic discipline, with four scientists who wanted to create systems that could replicate human thinking without using a programmer. In 1992/1993, there was a sudden growth in AI. This led to an enormous amount of data.
Where to apply Artificial Intelligence?
So, how can AI be used in general? First, for complex tasks. Think about copious amounts of data that cannot be understood with the human eye or with an excel sheet. Second, dual and repetitive tasks – like copying information from one system to another. To do this, AI uses text recognition and image recognition and can automate this with fewer errors than humans. Third, AI can be used for dirty tasks. With image recognition, X-ray cameras can see things that the human eye can’t see. Fourth, AI is excellent for dangerous jobs. Instead of sending a person, smart cameras can be used.
Autonomous systems
Fast forward to 2022. AI is no longer ‘just’ the automation of the human brain. We started to create autonomous systems that can be compared with the human body. Our sensors, like eyes, ears, and skin, are connected with our brain, which makes analyses. Next, the brain sends answers to our actors (arms and legs). AI has come to a point where we have digitalized the entire system: We use sensors and autonomous data streams, we send them straight to the algorithm, and from the AI system, we create autonomous environments. Does this mean that humans will be unnecessary in the future? No, as you will read in the rest of the article. But because the artificial brain has become a hidden technology, a deep understanding of how AI works – and does not work – to collaborate with these autonomous systems is needed.
Transform your organization into a data-driven organization
For e-commerce experts: AI will help you balance your two challenges: customer experience and product-centric organization. This balancing is difficult to achieve with transactional systems. Your following best action or your next best offer is a customer-centric decision. But the organization is product-centric, with thousands of products. Here is where AI can make the difference because AI looks at it as a mathematical matrix. The balance will be optimized from a matrix point of view and not from a line point of view.
Do not start with a big bang when your organization is ready to use AI. Instead, start small and work in an agile way. Work on tiny KPIs, like improving your online conversion rate. Ask yourself what data is needed for this, if the data is available and if the data quality is good enough. Based on that data, you create insights that you use in customer interactions. Next, evaluate the impact of your interactions and improve them. This is how you drive a data-driven or AI-driven organization.
Will e-commerce specialists become obsolete?
Algorithms and data can help your organization tremendously. Will e-commerce specialists still have a job some years from now? Absolutely. It is essential to combine specialists’ knowledge with data platforms. AI projects and e-commerce do not fail on data and technology. They fail on people, processes, and understanding of which translation must be made. To increase customer loyalty online, you need to offer the customer what he wants from a context and timing perspective. Only when you get both correct will the customer accept it. What the proper context and timing are, depends on the organization. A great example of an AI-driven organization is Isero. Just read their story "Data, algorithms, and traditional product expertise: The right balance will boost your e-commerce".
When does AI collaborate with people?
Online click behavior is the new data stream on which AI brings you value. In online click behavior, you see the psychological state of a person. When you conclude from data, for example, that someone will buy anyway, you can adjust pricing. Location data plus online behavior data is worth gold for AI and for you to think about the proper interactions. The techniques for seeing value in it have been around for over thirty years. The shopping basket analysis was done 20 years ago. What has changed is the computing power to generate an immense amount of data while the customer is still on your website. Data and AI give you the tools to take the right actions. The correct action on an e-commerce site ensures you understand why the person visits your website and what he is doing. Based on where the customer is in the journey, you will use AI personalization techniques. You can only make generic recommendations when you do not know someone and the context. You can personalize when you do not know the person but know the context.
The good, the bad, and the ugly
AI, however, is not the holy grail. An AI brain is trained in a particular context. If the context changes, you must retrain the brain. Algorithms must constantly be (re)trained based on the environment. When it comes to AI, looking at the good, the bad, and the ugly is essential.
The Good
The good thing is that we tried to replicate the human brain – and it does not work like this. Yes, AI is outperforming the human brain in terms of accuracy and processing data. The human brain, however, beats AI in terms of, e.g., adapting to change, physical collaboration, creativity, and emotions. However, the human brain and AI fail regarding diversity and inclusion topics. Human decisioning errors are in the data. Our systems are biased if we use that data to train our systems. It is good to realize this. Also, it can be used effectively. Some companies use AI to see which managers are biased, for example.
The Bad
The bad is that business has not done a lot to understand how AI works. Many organizations focus on collecting data, but having data is not enough to use them well. 80 percent of the time in your AI project, you will spend cleaning data, ensuring data is complete, and understanding your data. This asks for hiring data engineers when you start an AI project. Of the 85% of the companies that begin AI projects, only 25% can put AI into operations. This is because of understanding the importance of internal people and processes to run AI.
The Ugly
The ugly is that we live in a world without regulations for AI. Data seems to be the holy grail for many companies. So, AI engineers created autogenerated algorithms: algorithms that form data. This has led to the rise of companies that create deepfakes and other fraudulent deeds. The good news is that Europe has made the European AI Act. It will be effective shortly and asks companies which data they use, who trained the system, how they retrain the system, and so on. This Act was created to recreate the trust in AI systems for the coming time.
In short, companies must take on full responsibility and accountability when using AI. Make sure that everybody in the organization gets acquainted with AI from a certain level from a multidisciplinary point of view. AI is still a very narrow technology that is not entirely ready yet to be integrated into our more complex world. The role of the e-commerce specialist will be to use social and collaborative skills to make sure human intelligence and AI systems go well together.
Tips when you consider working with AI:
-
When you want to start using AI, do not start with a big bang. Instead, start small: work on tiny KPI’s. Work agile, and do not forget to evaluate and improve based on your evaluation.
-
Use AI to see correlations in giant data sets, but use your human brain to see if there is causality or not. AI cannot reason what we do with our frontal cortex. AI is the thinking fast part. The slow part is a company's strategy, where the objectives and KPIs are defined.
-
Hire data engineers when you start an AI project.
-
Evaluate feedback from your AI system, adapt your data, and retrain your model before moving on to the next step.
-
You can use pre-trained models as a starting point, but they might not be working for your context.
-
Never forget that you can only get the most out of AI if you combine Artificial Intelligence with human intelligence.
-
Use AI responsibly.
Are you interested to see the presentation of Mieke or other inspirational videos of B2B digital experts? You will find them by clicking this link.
