Book Summary

—

Today's Amazon Deals - https://amzn.to/3FeoGyg

—--

Book Summary

Understanding AI

Artificial intelligence is something of an umbrella term. It can be used to describe a range of different technologies and programs used for a variety of functions. On a small scale, one form of AI is a system that automates responses to customer queries. On a larger scale, you have robots that can carry out full conversations. Understanding the differences between the different technologies and how they can fulfill your goals should be your first step in planning an AI solution.

“Understanding the differences between the different technologies and how they can fulfill your goals should be your first step in planning an AI solution.”

AI Capabilities

According to the author, to better understand AI, you need to think more broadly about the topic. Don’t just think about specific AI methods or solutions but also consider what the end result of an AI solution would be. In other words, what the AI is actually doing or what it is capable of doing.

Davenport looks at the main three capabilities that most solutions provide:

Automation: AI is often used to complete structured and repetitive tasks. This can be advantageous; after all, machines can usually complete these types of tasks much faster than a human, and with more accuracy.

Insight: Many AI solutions are used to gain insight into different areas that might be impossible or painstaking for a human to achieve. For instance, machine learning algorithms can comb through thousands of records, perform calculations, and extract meaningful insights from data.

Engagement: AI can be used to engage with customers and employees without the need for human contact. This is similar to automation but it relies more on human oversight and has to be more complex to handle a wider variety of situations.

Some AI solutions offer a combination of automation, insight, and engagement. For instance, you could have an AI “help desk” that can engage and answer questions and then automate basic tasks based on what a customer requires.

AI Basics

Davenport notes that, contrary to popular belief, most of the AI solutions available today aren’t the flashy kind we often see in the movies. Rather, they’re more practical and can offer businesses untold benefits. Many industry leaders, such as Facebook, Amazon, and Google are already using AI in different capacities, for search suggestions or identifying customer buying patterns, for instance. Moreover, many companies are now creating ready-made AI solutions for purchase, making it easier to get into the field without having to develop a solution yourself.

To fully comprehend what AI is and how it can help you, you’re first going to need to understand the different types of AI that are available. The AI Advantage provides a breakdown of the main AI in use today:

Statistical machine learning uses complex algorithms to discern patterns from data. This can be used, for example, to identify or classify customers based on a list of attributes.

Deep learning uses algorithms based on the brain structure to recognize patterns. This can be used in image and voice recognition software to extract meaning.

Rules-based expert systems are a set of logical systems that follow a rule set given to them by human experts. These can be used by banks to provide customers with credit approval, for instance.

Robot process automation automates digital tasks and interfacing with systems. For instance, companies can automate processes such as replacing credit cards or validating online credentials.

Davenport notes that it’s important to dig deeper into the subject if you want to fully grasp what is out there and what is possible when using AI.

Developing a Strategy

Once you understand what AI is and what it can do for you, it’s time to start developing your strategy. For instance, you need to decide what type of solution you require and how it will integrate into your current environment. You also need to consider aspects such as cost and the benefits you hope to derive from whichever solution you opt for.

Developing Strategic Objectives

A good place to start is to identify what your aim is. According to the author, you need to figure out what problem you’re trying to fix or what area you’re trying to improve. Once you’ve identified the outcome you wish to achieve, you can make decisions about what solution you require or if it’s even feasible to undertake.

What if you’re having trouble deciding between the different areas? Davenport suggests taking a look at the situation from your customer’s perspective. So, which changes will most benefit the customer? You might be quick to jump to a solution, such as a chat bot that can help customers with frequently asked questions. But these aren’t perfect and can even deter customers, who may prefer engaging with a real person.

The author also advises you to put some thought into whether you need an internal or external change. An external change, like a chat bot, can be impressive but might not be that helpful unless done well. Some internal automation, on the other hand, may never be noticed by a customer but might speed up your entire process—and, ultimately, enhance the customer experience.

“Some internal automation, on the other hand, may never be noticed by a customer but might speed up your entire process—and, ultimately, enhance the customer experience.”

Aligning Your Strategies

When developing your AI strategy, you need to consider more than just the technologies you’ll use. Davenport notes that there are other factors that may determine whether your solution is a success. The most obvious consideration is cost. Remember, many good solutions will be expensive—and that’s without adding the hidden costs of implementation and time to transfer. The author highlights some other key areas that should be aligned properly for a successful AI solution:

Content: Most AI solutions require some kind of content to run on or learn from. For instance, customer data or large galleries of images. In some situations, you can use whatever information you already have stored. In other situations, you might need to start collecting the right data or consider outsourcing it.

Talent: Whatever solution you opt for, you need to have experts on hand to work on it and work through any issues that may arise. This is especially true if you plan on developing a solution in house. Of course, experts can also be outsourced.

Partnering and acquisition: If you do lack the talent in your own company to develop a strategy, it may be worth considering partnering with another company to develop this solution. Many bigger companies, which are not IT or AI focused, have also elected to acquire smaller AI-focused companies to develop solutions for them.

Ambition: Keep your ambitions realistic, especially if you’re new to the world of AI solutions. Radical transformations rarely work out so try to keep your projects small. Stay within your own realm of business as well. Start with smaller projects and grow from there.

Once you’ve developed a strategy, you can determine if the solution is viable. Davenport notes that you may find that the solution you’ve identified isn’t the right fit at that point. If so, it’s better to hold off until you’re ready.

Managing AI Responsibly

AI represents an exciting new world of possibilities but it’s not without its drawbacks. For instance, you need to consider its indirect impacts, such as how a new solution might affect an employee’s workload or responsibilities. AI can have many shortcomings if not managed properly. You will need to prepare for the possibility of setbacks on top of dealing with the change management that comes along with new AI solutions.

“You will need to prepare for the possibility of setbacks on top of dealing with the change management that comes along with new AI solutions.”

The Drawbacks of AI

Davenport acknowledges that although AI can be beneficial, it’s not without its shortcomings, especially in companies that rely on a lot of interaction with customers or users. AI isn’t perfect and you need to be careful, especially when handling things like sensitive customer or user information. The author also notes that many companies have had problems with putting too much faith in their AI systems. They don’t oversee their systems properly and problems arise when everything is left up to the machine. Ultimately, there’s still need for human input and human oversight.

Many machine learning solutions do their learning in an autonomous manner; so much so that sometimes it’s impossible for a human to understand how it comes to certain conclusions. Davenport notes that this isn’t a problem, in most cases, as long as it reaches the correct conclusions. However, sometimes AI can unintentionally develop a bias, and this can be damaging. In other words, the program learns to discriminate against certain users or groups, even though that was never the intention. In these situations, it’s important to be transparent with customers, stakeholders or end-users about what is going on and work to rectify the situation.

Change Management

To implement any AI solution successfully, you’ll need to perform change management. The AI Advantage explores the need to get everyone on board with the idea of a new technology. As the author observes, this is vital in order for the new solution to work. For instance, you should collaborate with employees, executives, and any other stakeholders who may be able to offer input into the types of solutions that are required.

“The AI Advantage explores the need to get everyone on board with the idea of a new technology. As the author observes, this is vital in order for the new solution to work.”

Of course, not everyone will be keen to embrace AI. Davenport notes that a common misconception is that AI will make humans redundant in the workplace. But this isn’t usually the case. If fact, AI usually only fills one task and doesn’t replace the entire job. AI is also not fool proof; there’s room for error, which means human oversight is needed to ensure optimum performance. So, in this way, AI is useful to supplement jobs, not replace them—something you should keep in mind when developing a viable solution.

Ultimately, you need to keep an open mind and work with the people around you to see how this new AI future can work for everyone. For instance, it could mean reclassifying an old job or focusing the job into more niche areas that the AI can’t handle.

Finally, Davenport notes that while AI initiatives might seem bold, new and exciting, the changes don’t hit people equally. For a successful transition to a new initiative, there are three main types of employees who are most important and who you should try to keep happy:

Experts: In order for your AI initiatives to work, you’ll require assistance from employees with knowledge in the area. They’re usually the ones training and maintaining the technology. With the new system in place, some of their previous tasks may be supplanted and they may feel put out by the changes. So, make sure they know they’re working with the new systems, not being replaced.

Decision makers: You are likely implementing AI systems to improve decision making, which can feel threatening to employees who have previously been responsible for these tasks. Again, stress that these changes are meant to augment their jobs, make them easier, rather than replace them.

Recalcitrant learners: Some people will be slow to adopt new things and others unwilling to learn about these new technologies. Make sure everyone’s on board and positive about the change; this makes them more likely to adopt the changes quickly.

Regardless of who you’re focusing on, it’s important to stress the benefits of the solution and work with the people involved to figure out how they can integrate and utilize the solutions to improve their work.