H&R Block Tax Consulting (IBM Watson)

I have yet to meet a person who enjoys filing income tax returns every year. 140m US citizens file tax returns every year. Most need professional help from tax consultants. H&R BLOCK is one of the largest tax consultants in the USA with over 10,000 offices filing tax returns for 11 million Americans. They employ 70,000 tax experts for preparing tax returns. H&R BLOCK has been losing business in the last few years to online tax programs like TurboTax, hence were seeking a strategy that would craft a more engaging, interactive and rewarding experience for clients. HR Block CEO Bill Cobb’s goal was to “reinvent the retail experience of taxpayers.” The solution? Use Artificial Intelligence as a tax specialist to accompany and assist the HR Block representative in the process of preparing tax returns together with their clients.

H&R Block opted for IBM Watson as their preferred AI platform. To train the Watson AI system as a tax specialist: it was fed with the enormous 74,000 pages of Federal tax code and thousands of pages of tax law changes every year. This is a vast amount of information to take in, sort through, and decide how it applies to each customer’s tax situation, depending on issues such as marital status, job losses, dependents, mortgages, and capital gains, but Watson does this with ease. “This speaks to Watson’s strengths as a learning machine,” said Rob Enderle, an analyst with the Enderle Group. “Unless you are an incredibly unusual person, there is no way you can know how to optimize a tax payment or return absolutely. There is simply too much information to learn and so much new [data] being created that even if you were able to read and understand it all, by the time you were done you’d still be terribly out of date.” Since Watson can collect and categorize data, it can absorb new tax information in real time, keeping it up to date with new laws and adjustments to old laws, Enderle said. In addition, the best tax experts at H&R Block, train Watson on the proper response to thousands of customer questions that have cropped up in the past years. As a result, Watson becomes a central repository of tax laws, top expert knowledge, and the history of all questions and responses. This knowledge and expertise becomes instantly accessible to all tax consultants of H&R Block. “Once [Watson is] fully trained, H&R Block should be able to show absolute proof that their customers pay less taxes because the system will be able to generate reports that accurately showcase this result in aggregate,” Enderle added.
(http://www.computerworld.com/article/3173283/artificial-intelligence/hr-block-turns-to-ai-to-tackle-your-tax-return.html)

When a client sits down with an H&R Block tax preparer, the preparer asks questions that are also provided to Watson. This information involves tax-relevant events that happened to the customer during the previous tax year – home purchases, marriage, birth of a child, a child entering school or a family member leaving the military. Watson uses that information to draw insights into the tax implications of those new jobs, home sales, and marriages and offers recommendations. During the session, the customer follows along on a monitor as his or her taxes are prepared, and Watson suggests different possibilities. All decisions are taken by humans ?—? the tax preparer and the taxpayer ?—? but guided by the expertise and accuracy of Watson AI. Most customers find the process not just very interesting, but at the same time rewarding since they generally receive higher refunds resulting from lower tax liabilities because of Watson AI’s thoroughness and accuracy. Human tax consultants are able to secure refunds for only about 75% people who file taxes. H&R Block with Watson AI promises refunds for up to 85% of customers. Customer return rate is another benchmark for H&R Block. They aim to raise the return rate from 75% to 80%.

Some key insights from this example:

Expert Knowledge Aggregation – Learning, classification, referencing and storage of massive amounts of data is a unique ability of Watson AI systems. Stored data can be continuously and immediately updated. This allows AI systems to draw from a vast range of up-to-date knowledge for each case. The system is ever evolving and learns and updates itself with every new case.

AI Recommends and Human Decides – We humans are not yet used to taking advice from machines in critical personal areas such as taxes, health and law. Advice from humans is more comforting and trustworthy, even if humans are not as knowledgeable as machines. We feel better hearing about a diagnosis from a human doctor than from a machine, even if the doctor made the diagnosis based on the results from a machine. Watson doesn’t cut out humans completely. It aids the process and lets humans do their job better.

AI Platform – a Fast-track to Results – AI platforms provide a shortcut to benefiting from AI technology. Amazon, IBM, Google, and Microsoft offer AI platforms for quick development of solutions, without the need to hire many AI technology specialists. H&R Block remains focused on its core business of tax consulting and just simply leverages the IBM Watson platform for AI. This gave them the first-mover advantage in offering the benefits of AI to tax-paying customers seeking better service and rewards.

Big Data is Heaven for AI, Hell for Humans – As discussed in the first part of this book, too much data overwhelms us and leaves us disoriented and confused resulting in decision paralysis or just plain poor decisions. In contrast, AI relishes data. Its decision making quality and accuracy improves with additional data. Massive tax codes and constant updates are handled by AI with ease.

Salesforce.com (Einstein)

“We cannot solve our problems with the same kind of thinking that we used when we created them,” Albert Einstein.

Fig. 9.2 Einstein from Salesforce applies AI to customer data

With this statement in mind Salesforce set out to rework its popular suite of customer facing cloud-based business products for companies, based on AI called Einstein. After a long planning period and the strategic investments in AI talent and companies, the AI enhanced suite was released in early 2017. You, as a business person, might know the challenges with CRM systems and other customer facing tools in service and marketing. Today’s tools give you a means to manage your data but the true value of this information lies between the lines and has to be extracted from the figures. You have to build models to understand the data and act upon it in such a way that you can leverage its full potential. Most implementations we have seen in our careers do not even come close to utilizing the potential. People get bogged down in transferring the data into the system, keeping it up to date and relevant and doing mass actions to act on it. But, since AI is good at understanding data, and detecting trends, associations and causes, it can prepare the right decisions and allow you to act or even takes control of acting with the right response at the right moment. Some new areas are discovering new insights about your customers like who are your best customers, predicting outcomes to make smarter decisions like what offer to send to customers and identifying which customers are considering leaving you. It recommends the next sales, service, and marketing actions.

By integrating AI in its suite of cloud-based applications, Einstein is removing the complexity of AI and enabling any company to deliver smarter, personalized and more predictive customer experiences. For example the Salesforce CRM Software delivers automated customer insights to its user based on AI. Powered by advanced machine learning, deep learning, predictive analytics, natural language processing and smart data discovery, Salesforce claims that Einstein’s models will be automatically customized for every single customer. It will learn, self-tune, and get smarter with every interaction and additional piece of data. And this intelligence is embedded directly into the context of your business data.

Salesforce’s cloud-first strategy and experience and understanding of ecosystems plays nicely in leveraging AI on a bigger scale. The Einstein API allows 3rd party developers to create their own apps, utilizing the API resources and integrating with customer data. One example of integrating AI-based vision with AI-based customer preference can be taken from real estate app “The Dreamhouse”. Customers are shown real estate images, compiled by Einstein’s AI from a pool of offerings, based on their profiles behavior preferences built by the CRM AI. Sound impressive? It even improves over time. If Einstein is anywhere near as useful as Salesforce claims, the technology will supplant some human workers ?— maybe a lot of them.

Some key insights from this example:

Leverage the value of data – and CRM systems are likely to have lots of data. Instead of people wasting time trying to make sense of it and acting upon it, let AI do the work. It will be a lot faster and more thorough.

Learns and improves over time – customer data grows over time, in fact most of the insights come during the customer life cycle. AI continues to learn and improve its learning to better suit your customers preferences.

Improve all aspects – once you have mastered the AI technology it allows you to improve all aspects of your offering. Salesforce has integrated Einstein AI into all of its cloud offerings thus improving the way it works and adding new AI-based capabilities.

Build a platform – Einstein AI enhances the Salesforce offering, making it more valuable to its customers and allowing Salesforce to charge a premium price. By integrating additional data sources and providing an API to external sources it increases its value exponentially, while the competition is struggling and falling behind.

Leverage 3rd party developers – One company cannot create all value scenarios and apps for such diverse industries. By providing access to Einstein Services like image recognition and coupling those with Salesforce-based data services, Salesforce customers get a richer set of software services without requiring development by its customers.

Tesla Autonomous Vehicles/Mobility (diverse areas)

“Full autonomy is going to come a hell of a lot faster than anyone thinks it will and I think what we’ve got under development is going to blow people’s minds,” said Elon Musk, CEO of Tesla, in a press release press release in August 2016.

Fig. 9.1 Tesla, prime example of sensor based autonomous driving with AI

Did you know Tesla has never turned an annual profit and is not predicted to do so until 2020. So how does this firm have a market value on par with some of the world’s largest automobile manufacturers? Tesla is not just an automaker, but also a technology and design company with a focus on energy innovation. By using the latest computing technologies to bring new benefits to its customers, like autonomous driving and related services, it has disrupted the market.

Autonomous cars are vehicles capable of sensing the environment surrounding and navigating without human input. To achieve this, Tesla first used cameras and radars to detect and understand obstacles on the road to help steer its autonomous vehicles. The problem that emerged however is that sometimes these sensors could be tricked by complex and seemingly illogical situations. For example, a low bridge with a dip underneath it may appear to be a solid roadblock from far away (this also happens to the human perception, by the way). In response to this, Tesla began collecting data from other Tesla vehicles to observe how human drivers behaved at a particular location. If other vehicles did not brake consistently, then no braking would be required for autonomous driving. Similarly, when the observing AI noticed drivers slowing at certain locations, the AI implemented mild breaking for autonomous vehicles. These decisions are combined to data from sensors to formulate the best decision for any given situation. In doing so, Tesla built a map based on artificial intelligence, and is employing it as an input for all its cars in combination with the radar. It even works in very difficult visibility situations, providing additional safety for drivers and their cars.

For all of this Tesla is using artificial intelligence algorithms which also define the confidence levels. This means if there is a 99.99 percent confidence level that there is an obstruction, it will add a full break for that designated location on the map and combine that with the sensor input, at any given time.

Tesla estimates that, after 10 billion kilometers are recorded/combined from Tesla cars, their car will be 10 times safer than regular human driven cars. By driving this distance it helps the AI to learn and find the best decision for any given moment. The advantage of this self-learning method is that it can resolve some of the most complicated situations for the sensors of cameras and radars. The more drivers will use it, the smarter it will become. This combination of learning from observing behavior (of drivers) from a wide range of drivers and combining it with input from sensors of the local car illustrates the power of AI for autonomous vehicles. It can overcome complex hurdles in discovering and reacting to obstacles on the road based on sensory data streams in combination with behavioral pattern. There are many similar approaches taken in Internet of Things solutions worldwide.

Some key insights from this example:

CEO is driving it – Elon Musk is the driving force behind Tesla’s advancement in Artificial Intelligence. With his wiz kid science background, passion and engagements in leading AI initiatives like OPEN AI, Musk makes AI the center of his work. This makes it clear for the organization that AI is not just a nice to have technology but a key building block of the future of the company. Elon Musk makes it a top priority and makes sure all understand.

It’s a way to outcompete other players – By leveraging AI, Tesla is able to achieve new levels of safety compared to today’s human driven vehicles. While many competitors are playing catch-up by adding sensors to their vehicles to allow them to support drivers in certain situations, Tesla is fully committed to letting AI deliver self-learning autonomous driving.

It provides true value – Next to improved safety, Tesla’s AI-based autonomous driving offers a new level of comfort and frees-up personal time for its customers and, in doing so, creates a new value proposition. In addition, autonomous vehicles create even more value for customers by exercising the capacity to perform other services while not being in use.

Its value improves over time – Tesla vehicles are known to improve over time, just as smartphones improve with new app capabilities, Tesla vehicles are updated continuously. AI means their capabilities keep on improving, thus making them more valuable.