Introduction: Welcome to a New Paradigm
Artificial Intelligence has come to corporate America. In fact, a recent study found that EVERY company on the FORTUNE 50 is already using artificial intelligence to help teams work more joyously, automatically, and efficiently.
See, AI strips the monotony out of work-life, and as AI development costs continue to decline, the technology is becoming increasingly accessible to companies of all sizes.
This guide represents everything we've learned about bringing AI to the enterprise. Not only does it cover everything you need to know about AI and its capabilities, we'll also share our lessons to smoothly deploying AI across your company.
To be sure, AI represents a big shift for your organization and so we wrote this guide so that everyone from your leadership to your IT team to your workers can prepare.
All the information you need to plan and execute your transition to AI is contained within these pages. Hope you enjoy!
Chapter 1.
What is Artificial Intelligence?
Artificial intelligence is a software capability that allows computers to detect patterns and abstract rules based on a series of extremely complex inputs.
These inputs can include any dataset, including:
- pixels in an image,
- sound waves in an audio file,
- letters on a page,
- the mathematical relationships between all the words in the English language,
- financial transactions of the Fortune 500,
- the location of every star in the galaxy,
- etc.
The dataset itself doesn’t matter much. As long as an AI model can take a look at the input and match it to its output, the system itself can start to map — and ultimately predict — the rules that connect inputs and outputs.
For example, a financial institution may have a dataset of historical stock market prices and lots of information about each company. By matching company information as the input with stock price as the output, an AI could learn to identify company information that is highly correlated with a rising stock price.
What’s great about AI is that it is uniquely able to consider thousands of inputs and correlations far too subtle for humans to even notice. For example, almost no stockbroker integrates CEO commute time into their investment strategy. However, artificial intelligence has found it to be a relevant predictor of CEO performance.
In any case, over time, these predictions become more accurate. Eventually, the model becomes reliable and useful enough that it can be deployed and put to work.
Let’s look at another example.
At a factory, quality assurance officers must pull broken eggs from a conveyor belt. To accomplish this task, the inspectors will look for obvious indicators such as a cracked shell or a shimmer of yolk on the conveyor belt. In other words, these folks are using visual information (inputs) to categorize eggs (outputs).
This effort is a perfect task for an AI model. In fact, if we were to assign an AI model the same task, it would tackle the problem much in the same way as its human counterparts: it would look at an egg and check to see if there were any visual indications that it was cracked.
To accomplish this task, however, an AI would need to be trained. In this case, the model would need to chew on hundreds, maybe thousands, of photos of cracked and uncracked eggs to start to figure out the pattern.
Eventually, the model would create mathematical rules to define what a cracked egg looks like. With more training, the accuracy of the model would improve, and eventually, it would be reliable enough to put into production.
A Note About Accuracy and AI
Artificial intelligence is not yet capable of achieving 100% accuracy. So, depending on the use case, data scientists must make judgments about whether an AI model meets a certain threshold to put it into production.
Most AI models can get to about 75% accuracy without too much trouble. Beyond that, there’s a logarithmic increase in (a) the number of inputs the model needs to train on and (b) the amount of processing power it takes to create the model. The former is time-consuming, the latter is expensive, and in most cases, 100% accuracy isn’t necessary. Depending on the task at hand, the accuracy threshold will vary.
This lack of 100% accuracy is one of the reasons self-driving cars still have a long way to go. In data science, a 90% accuracy rate is quite good but imagine if an autonomous vehicle only noticed 90% of the pedestrians who crossed the street.
Moreover, how things are inaccurate can also be a factor. For example, in a lung cancer model, false positives aren’t nearly as bad as false negatives. Therefore, the best disease diagnostic models will be optimized to detect every single possible instance of cancer (avoiding false negatives) — even if it means a few healthy lungs get incorrectly flagged along the way (allowing for false positives).
Experts who build AI solutions must decide on acceptable accuracy thresholds before putting a model into production.
What does it mean
to put a model into production?
Well, as we’ve discussed, AI models are really good at matching inputs to outputs. However, for that capability to be useful to anyone, the model must be piped into some kind of app or hardware to be put to work.
In the egg example, the food processing plant would have to do more than train the broken-or-not-broken model to get any use out of it. It would need to set up a camera to film the eggs coming down the belt, pipe that footage into the model for processing, and then alert the line workers of a broken egg via some kind of interface.
Most AI service providers consider training a model and putting it into production as two separate steps. Because the former is much more complex (and expensive), many IT departments opt to do the production piece themselves. In most cases, the average programmer can access the AI model (usually via an API) and put it to work.
What is Deep
Learning?
For AI, identifying whether an egg is broken is relatively simple. By contrast, looking ahead at a red, round object and labeling it a “stop sign” is much more complex.
This is where deep learning comes in. Deep learning enables AI models to sort things into tinier and tinier categories. AI experts think about this sorting as a layered process.
In the case of the stop sign, the relevant layers might be:
- Distance from car.
- Shape.
- Estimated height.
- Position on the road.
- Color.
- Text on the sign.
- Etc.
To identify the stop sign, a self-driving car would have to pass the image of the stop sign through a series of layers to make sure it wasn’t a mountain, a tree, a speed limit sign, or a Starbucks. By layering simple AI categorization models on top of each other, AIs become much more sophisticated.
Let’s look at another example related to emotion detection. Could an AI detect whether someone was in a bad mood? To consider that question, let’s first think about how humans can tell whether someone is in a bad mood.
In order to make that judgment, our brains must integrate a lot of visual and auditory information about the subject.
For example:
- Mouth position (smiling, frowning)
- Eyebrow position (raised, narrow)
- Pronounced wrinkles (laugh or frown lines)
- Tone and volume of voice (yelling)
- Speaking speed
- Posture
- Shoulder position
- etc...
In fact, there are so many subtle and not-so-subtle indicators that it’s nearly impossible to list them all out. What’s nice about AI is that we never have to define those messy and long-tail indicators. AI’s superpower is that it can figure out the rules for itself.
By showing an AI model dozens of faces and labeling each with an emotion, the model could be trained to find the pattern, thereby integrating a wealth of obscure information about how someone looks to predict how someone is feeling.
This is the same facial recognition capability that Apple uses to unlock your iPhone. In that case, instead of matching your face with an emotion, the system is ensuring your face matches your account.
AI researchers are using images of people’s faces as inputs for all kinds of models, including lie detection, age detection, disease diagnostics, and even imaging technologies that age-up missing children.
As you begin to think more and more about the ways artificial intelligence can transform your company, challenge yourself to think in terms of inputs (data) and outputs (categorization).
Remember, AI is amazingly powerful. AI models can use Xrays to diagnose disease, billions of financial transactions to detect fraud, and boxes of handwritten case files to prove someone is innocent.
Don’t worry if the distance between inputs and outputs seems vast. AI is more than capable of making those connections.
Chapter 2.
What are the Major Types of Artificial Intelligence?
So now that we've explored how AI is a tool for finding the relationship between inputs in outputs, let's define what some of those major inputs and outputs can be.
Inputs
Semantic Data
By now, AI is super-human in its ability to understand and process language. This is obviously super useful because a huge part of the world’s information is one kind of communication or another. We’re talking documents, books, handwriting, articles, speech, etc.
Natural Language Processing (NLP) models think about language in terms of each word’s relationship to related words. We can think of NLP models as a vast 3-D word cloud where related words cluster together forming vectors of associations and understanding. In effect, language processing models convert language into mathematical relationships.
Although this may sound a bit abstract, our brains actually process language in a similar way. Consider this sentence:
The dove dove into the water.
Most of us can pretty easily intuit from the sentence structure that the first dove is the noun and the second dove is an action related to water.
Languages like German, Chinese, and Korean drop these associations right into their vocabulary. For example, in German, kindergarten literally means child garden. In Korean, the word for fish (물고기) is a compound of water (물) and meat (고기).
All of this is to say that by quantifying the relationships words have to other words, AI language models have gotten extremely adept at learning to read, write, translate and summarize with ease.
Quantitative Data
AI models are built on complex mathematical equations so it shouldn’t surprise you that AI speaks the language of numbers. AI models can vacuum up all sorts of quantitative data such as statistics, analytics, dates, sequences, financial information, etc.
Multimedia
Whereas semantic and quantitative data uses abstract concepts as inputs, multimedia data is much more concrete. When considering what multimedia data you might be able to use, think about your senses:
- Visual data can include photography, video, satellite footage, video surveillance
- Auditory data such as sound recordings or RADAR.
- Sensory data such as temperature or LIDAR.
Hell, if you’ve got an electronic nose, you could even consider olfactory information.
By the way: data scientists can build AI models that consider datasets from all of these categories at once. For example, you could build a healthcare diagnostic model by integrating a doctor’s description of the patient’s symptoms (semantic), their bloodwork (quantitative), and chest Xrays (multimedia).
Outputs
As we discussed in Chapter 1, AI models build intuition about the relationship between inputs and outputs. To better understand how your organization could put AI to work, you’ve got to consider both. Broadly speaking, AI can be used to do two main things: surface information or create content.
Surfacing information includes things like categorization, pattern recognition, and prediction. These outputs can then be piped into systems that either (a) deliver this essential information to stakeholders for better decision-making and/or (b) put this information to work via some kind of hardware or software.
For example, an AI model that surfaces a stock prediction could be put into production as a daily email that arrives in your inbox or as an app that will automatically conduct trades for you.
Alternately, AIs can be used to create content. In a broad sense, the types of content that cognitive intelligence can create is similar to the inputs that AI can consider. For example, AIs can produce both articles and images of all kinds of crazy iterations, as we’ll discuss below.
Let’s consider each of these outputs in more detail:
Surfacing Information
Categorization
AI is all about using data to make judgments or find patterns. In the case of categorization, AI creates a series of mathematical relationships to map the relative similarities and differences of similar objects. This kind of labeling is useful for all kinds of businesses.
Consider these examples:
- Quality assurance inspectors determine if a product is defective.
- Doctors determine if a patient’s bloodwork means they are healthy.
- Mortgage lenders determine if an applicant is highly qualified.
- Grocery stores determine if their produce is starting to rot.
- Marketers identify conversion-optimized landing pages.
- Programmers locate broken code.
With enough examples, an AI model could be trained to support the labeling/categorization process of any of these use cases.
Pattern Recognition
We can think of pattern recognition as a much more complex version of categorization. Whereas categorization usually considers one set of inputs, pattern recognition can consider an almost limitless number.
It’s this type of AI capability that enables credit card companies to detect fraud, retail organizations to elastically price items, and self-driving cars to make instant driving decisions.
When imagining how to bring pattern recognition capabilities to your organization, think in terms of the systems or processes you have in place to make decisions.
Predictive Analytics
With the right data, artificial intelligence is capable of making all kinds of predictions from heart attacks to stock market fluctuations to hurricanes. Very few companies are tapping into this future-facing capability and those that do tend to be primarily focused on tactical use cases like predictive maintenance.
However, having visibility of future events means that business leaders have the tools they need to make adjustments to their supply chain, their strategy, and the execution of their objectives.
With the right data, you can answer almost any question about the future. The oracle is real.
Personalization
Today AI-driven recommendation engines power your Netflix queue, your Amazon home screen, your Google search results, and your Spotify playlist. Every single time you read or click-through (or click away) these companies are gathering information about your interests and your preferences.
In fact, it is a similar engine that is driving much of the advertising across the internet. You know that feeling you sometimes get that Facebook is listening to you? The reason this may feel so is that their predictive engines are so powerful they’ve mapped a veritable galaxy of data points to identify things that they should try to sell to you.
Personalization offers companies the opportunity to have perfect product-market alignment. By being able to offer each of your customers something a little different, you ensure that the products and services you offer resonate in an unforgettable way.
Creating Content
In a broad sense, all of the inputs we discussed above can also become outputs for an AI model so let’s take them one at a time. For example:
Written or Spoken Communication
AI can generate all kinds of language as outputs, such as articles, summaries, labels, etc. We’re not just talking printed text here. As Amazon Alexa has shown, AI-generated communication can also be piped to voice-activated assistants.
In the case of articles, AI has taken all sorts of information to auto-generate articles. Publishers are using the capability to automatically write weather reports, sports stories, and updates on financial markets.
Text summarization is particularly powerful and can be a huge opportunity for your organization. AI can literally take a 2,000-page patient case file, extract the necessary information, and summarize it into a paragraph or two. A use case might be a law firm that converts boxes of handwritten evidence into an index of every item that’s there.
Multimedia
Already AI is generating a wealth of visual or auditory content. Artificial intelligence has been used to generate images of human faces, color black and white photos, write music, and create soundscapes, produce charts and graphs.
Translating Inputs and Outputs
As a final note, keep in mind that AI models become crazy powerful when you use them to translate one input type into a different output type. For example, an AI model could be trained to convert drone footage of a disaster area into a 12-page summary report for FEMA.
To be sure, building such a model would be complex, requiring both image processing layers to determine damage and text summarization layers to write that damage into easy-to-understand paragraphs — however, it is this exact type of complexity you should be thinking about when imaging the kinds of AI solutions to bring to your company.
Remember: artificial intelligence can transform information into almost anything.
Chapter 3.
What are some examples of Artificial Intelligence?
In this section, we’re going to get into the specifics of AI and cover some use cases that are being used across a variety of sectors, including Finance, Hospitality, Healthcare, Manufacturing, Product Development, and Smart City Infrastructure.
Natural Language Processing
Natural Language Processing allows computers to make sense of all kinds of communication, from handwritten chicken scratch to speech. This capability has given rise to all sorts of technologies from better spam filters to SIRI.
Finance. AI can help financial institutions bring NLP to large volumes of text and speech data to extract information, gain insights, and streamline manual tasks. Consider bringing automated summarization to legal documents, earnings reports, or job applications.
Hospitality. Interactive chatbots can create custom itineraries and book multiple points of travel. With natural language processing, “I want to go to Paris.” can lead to your customers booking air travel, lodging, restaurant reservations, and hotel tickets all from a single interface.
Healthcare. Natural Language Process can bring automation and summarization to the process of sharing, evaluating, and summarizing patient case files. (See our related case study).
Product Development. Natural Language Processing can support faster prototyping, intelligent programming assistants to make it easier for your team to write code, and machine learning to automatically refactor code.
Retail. Natural Language Processing can help your customers find exactly what they are looking for by speaking to an AI agent in a real way. Suddenly, “I’m looking for a wedding gift for my brother” becomes a back and forth conversation to give the buyer exactly what they want.
Smart City Government. NLP agents can comb social media and automatically aggregate it into actionable insights about how to improve your city. For example, multiple tweets complaining about potholes could surface that as a high-priority item for city planners.
Image Processing
Image processing enables computers to make sense of visual information and detect patterns therein. When thinking about image processing, consider all kinds of footage from photos taken by camera to footage taken by drone. Furthermore, because AIs aren't limited to the visible light spectrum, computers can bring intelligence to thermal imaging, LIDAR, etc.
Finance. Use image processing to keep tabs on markets of importance. For example, monitor shipping docks to keep tabs on supply chains, Walmart parking lots to anticipate fluctuations in retail demand, or oil wells to anticipate commodities futures.
Hospitality. In the Doordash age, how dishes photograph can be just as important as how dishes taste. With image processing technologies, restaurants can optimize plating to ensure that their most profitable dishes are getting the most attention. Similarly, hotels can use image processing to optimize the design, layouts, and promotional photography of their properties.
Healthcare. Use image processing on x-rays to detect disease. So far, AI has been used to detect breast cancer, early-stage Alzheimer’s, pneumonia, eye diseases, bacterial meningitis, and lots of others.
Product Development. By photographing items prone to design errors (like tires), product developers could train an image processing model to identify defects and predict recalls prior to going to production.
Manufacturing. Image processing can help you identify defects across the production line. By imbuing this system with artificial intelligence and self-learning capabilities manufacturers can save countless hours by drastically reducing false-positives and the hours required for quality control.
Retail. Intelligent Video Analytics can automatically and efficiently reduce shoplifting, predict burglaries before they happen, and improve loss prevention at point-of-sale — using the security cameras you already have.
Smart City Infrastructure. Most major cities have a vast network of live traffic feeds. By processing this footage using machine learning, city managers can not only respond to major traffic events automatically but also prevent them. When powered by AI, processing video means cities can have less congestion, better visibility on forthcoming disasters, and alerts for available street parking or expired meters.
Predictive Analytics
You can think of predictive analytics as the ability of an AI model to condense huge volumes of data into insights that people can use and understand.
Finance. Financial organizations have used predictive analytics to identify and target more profitable customers; better manage cash flow; anticipate demand fluctuations, and mitigate risk.
Hospitality. AI can use occupancy data, guest feedback, and self-reported guest data to gauge which upgrades or repairs should be implemented first and which improvements will deliver the best return.
Healthcare. Predictive analytics can be used in all sorts of ways to prevent undesirable outcomes. It can reduce patient no-shows, preempt 30-day hospital readmissions, predict resource allocation across divisions and service centers, and improve overall patient satisfaction.
Product Development. AI models can make accurate predictions about the scope and budget of your initiatives to limit cost overruns and schedule slips.
Smart City Infrastructure. Historical and geographical data can be put to work to predict where crimes are likely to take place. Such “pre-crime” initiatives have seen impressive results across big cities like LA, Chicago, and London.
Machine Learning
Machine learning is an automated process that AI algorithms use to integrate a wealth of information and turn them into action. In the case of predictive analytics, the integration of information leads to prediction. In machine learning, that integration can lead to all sorts of outputs as we’ll discuss below. Furthermore, machine learning uses recursive training to ensure that the conclusions its making are ever-more accurate and useful.
Finance. By analyzing billions of data points, fraud detection systems can actively learn and calibrate in response to new (or potential) security threats.
Hospitality. Most hotels know to adjust their pricing to adjust for seasonal fluctuations in demand; however, machine learning enables elastic pricing at all points of sale to ensure that both hotels and restaurants aren’t leaving a single dollar on the table.
Healthcare. Whereas disease detection typically relies on image recognition models, pharmaceutical research finds patterns in much more complex datasets. It’s unsurprising, then, that machine learning is being used at over 150 startups and 40 pharmaceutical companies to detect and deliver new molecules.
Manufacturing. Machine learning can transform a wealth of data coming from IoT enabled devices into action and automation. Machine learning systems can not only predict when manufacturing equipment needs to be pulled for maintenance but automatically reallocate other resources and lines to offset any losses in productivity.
Product Development. When integrated into the product development process, artificial intelligence makes everything, well, better. Software ships faster with fewer errors. Products leave the factory floor with better reliability and a lower risk of recall. Supply chain predictions are accurate and reliable.
Retail. Machine learning enables brick-and-mortar stores to get the same level of demographic and customer information as their eCommerce counterparts can. Learn who your customers are, how they move through your store, and how store layouts affect purchase decisions.
Greater Customer Personalization
In the AI space, we can think of greater customer personalization as one of the major capabilities enabled by machine learning. By integrating a galaxy of relevant information, companies are helping customers service the perfect recommendation at the perfect time.
Finance. Machine learning can help customers surface the perfect financial product or investment opportunity by integrating a wealth of information such as customer age, credit risk, etc.
Hospitality. Hotels and restaurants are starting to use AI to surface all sorts of thoughtful touches such as a guest’s preferred newspaper, toiletries, pillows, etc.
Healthcare. Healthcare models can transform a patient’s entire medical case file into a personalized health and treatment plan.
Manufacturing. Artificial intelligence is giving rise to all kinds of opportunities for personalized manufacturing. In one use case, companies are using AI to help their customers get custom-built clothing. In another, companies are using AI to produce artificial organs.
Retail. Customizations and recommendations across eCommerce are common but these technologies are coming to brick-and-mortar stores also. Through digital portals and interactive kiosks, AI can help retail stores hyper-customize their signage for each shopper that passes, integrating information about the weather, the shopper’s gender or emotional state, or fluctuations in the supply chain.
Smart City Infrastructure. AI is the master of processing routine requests and detecting abnormalities. Governments can exploit these capabilities to disseminate information, deal with citizen requests, or detect wasteful spending or fraud.
Chapter 4.
How Are Fortune 500 Companies Using AI?
Now that we've considered the broad ways that companies are using artificial intelligence to develop new products, optimize and automate repetitive processes, and provide better customer service, we thought it would be useful to see how some of the world’s biggest companies are using AI.
This collection is taken from our epic article: 50 AI Examples from the Fortune 500. Be sure to check it out for a deeper dive into any of these use cases.
A big takeaway from this list is that many, many companies are relying on outside resources to deliver their AI capabilities. This gives technology teams much more flexibility to deploy AI at scale.
In one of its first forays into AI, Amazon built a recommendation engine to make it easier for customers to surface more of the things they might like to buy.
Amerisourcebergen, the world’s most profitable pharmaceutical company, is bringing artificial intelligence to its benefit verification process.
Anthem embarked on a 12-month pilot project with the company, doc.ai to determine whether AI could be used to predict allergies.
Siri is a perfect example of how Apple runs on AI. The voice-powered assistant is designed for continual, at-the-edge improvement.
Archer Daniels Midland recently invested in a company that uses drones and image recognition technologies for farming.
By thinking creatively about opportunities for AI-powered automation, AT&T can deliver a better product for both its customers and its advertisers.
Bank of America is using AI to help reduce its labor force and drive more of its customers to receive help via automated systems and chatbots.
Berkshire Hathaway partnered with an AI data platform to optimize the insurance underwriting process.
Cardinal Health recently released a platform designed to support oncology professionals by making available a robust set of AI-powered capabilities.
In 2016, Chevron rolled out a machine learning system that could help it identify new well locations and stimulation candidates.
Citigroup launched an initiative to bring ML to the highly manual process of reviewing global trade transactions and ensuring their compliance.
Comcast is using AI to completely automate the process by which customers (a) receive phone support and (b) get technicians sent to their homes.
Since 2010, Costco has used the purchasing history of its 90 million customers as inputs to help them determine new store locations.
CVS Health partnered with AI startup, Buoy Health, to deliver AI-powered customizations and healthcare recommendations to their more than 1,100 Minuteclinics.
In 2018, Dell partnered with NVIDIA and Intel to launch a suite of hardware solutions designed to facilitate the deployment of artificial intelligence.
Dupont is using AI to scan their manufacturing equipment and ensure the quality of the products they produce.
ExxonMobil deployed an AI-powered algorithm to make it easier for its deepwater prospecting teams to drill at the bottom of the sea.
Fannie Mae partnered with Moogsoft to deploy the AI Ops platform across their organization for enterprise monitoring, reducing IT issues by a third.
FedEx is rolling out Roxo, an autonomous delivery robot that uses AI to navigate.
Ford was one of the first companies to deploy a neural net at scale and has since brought AI to both their assembly lines and in the operation of their sales departments.
Freddie Mac recently partnered with an AI fintech firm to bring ML & NLP to the mortgage underwriting process.
GE uses AI to cut the design process for jets and wind turbines in half.
General Motors deployed generative design to bring a 40% reduction in weight and a 20% increase in strength to their vehicle parts.
Last year, Google released Tensorflow, its open-source platform for machine learning, giving everyone access to one of the most advanced machine learning platforms ever created.
Home Depot deployed a monitoring system that uses prescriptive analytics to reduce shoplifting, employee theft, or other errors (known collectively in the retail industry as “shrink”).
IBM’s Watson is one of the most famous cognitive technologies in the world, winning Jeopardy, designing clothes, & inventing recipes.
Intel is looking to super-charge their data centers with more AI-powered capabilities. The shift to AI will affect 32% of its business.
Johnson & Johnson recently acquired Verb Surgical, a Google-backed startup focused on bringing robotics, data science, and automation to surgery.
JP Morgan Chase signed a 5-year contract with AI ad copywriters, Persado, increasing their click-through rate by 5x.
Kroger has an in-house team charged with framing, building, deploying machine learning solutions across their business regularly.
Lowe’s recently hired 2,000 tech workers, including a dedicated team of AI experts, to support their ecommerce and operational efforts.
Marathon has recently rolled out a suite of tools in its drive to automate its oil-drilling sites.
Metlife uses voice recognition to provide real-time feedback to its customer service team, offering on-screen advice like “finish your thought” and “you are speaking slower than usual.”
For several years, McKesson has maintained a partnership with a global professional services firm focused on bringing about digital transformation.
To monitor the quality of their potato chips, Pepsico bounces lasers off of them and uses AI to listen to the echoes.
Phillips 66 uses AI-powered robots to inspect oil tanks without having to drain them first.
In 2007, Proctor and Gamble released an AI-powered makeup quiz that could analyze your selfies to create a skincare regimen.
Prudential has made a big AI investment to ensure their customers are more precisely matched to their suite of insurance products.
Statefarm launched a $100 billion fund to invest in insurance startups that deploy AI and ML.
Target uses AI imaging technology to give customers virtual makeovers and drive makeup sales.
Like Amazon, United Healthcare is deploying natural language processing across large swaths of its business.
United Technologies recently deployed an AI-powered drone that can map convoluted and complex urban environments.
UPS’s On-road Integrated Optimization and Navigation system, uses AI to power one of the most advanced fleet routing systems on earth.
Verizon’s Connect is an AI-powered fleet management solution that can automatically process in-car telemetry and dashcam footage.
Walgreens partnered with Microsoft to develop new health care delivery models, technology and retail innovations to advance and improve the future of healthcare.
Walmart is bringing image processing to all of its locations to make it easier for employees to keep their stores running smoothly.
Wells Fargo has made a multi-billion-dollar investment in data analytics and is developing a variety of AI algorithms for their financial products.
Chapter 5.
Is my company ready for AI?
So, now that you’ve gotten a sense of what AI is capable of and how the AI development process works, the next thing to consider is whether your company is ready for an AI deployment.
To be sure, most businesses see a huge benefit to bringing AI to their organization but the path toward AI transformation might not always be clear.
Despite what can be an uncertain process, according to a Forbes survey, 84% of C-Suite executives believe AI is an essential component of achieving their revenue objectives. And while AI has historically been launched at big firms with large innovation budgets, AI is now trickling down to businesses of every size and composition.
Whether deploying a major AI initiative or a small pilot project, bringing AI to your business can represent a major shift for how your team does its work. Fortunately, the process can go much more smoothly if you take steps at the beginning to ensure an aligned transition.
In our experience, the best AI deployments come from multi-million-dollar companies that are (a) far along in their digital transformation and (b) have absolute buy-in from stakeholders throughout the business.
We’ll explore those pre-conditions below.
Digital Maturity
The greatest indicator that your company is ready for AI is the quality, accessibility, and structure of its data. In fact, for most deployments, the most challenging part of the process is ensuring that the data is organized and properly labeled so that an AI can develop an intuition about it.
Companies that are early in their data collection trajectory will be well-served by putting processes in place to collect it. Data scientists require thousands upon thousands of data points to successfully build AI models and so you want to make sure to start accumulating those reams of information now so you’ll have it if/when you turn to AI.
For those who are further along in their digital transformation, the challenge is to make sure your data is well organized, of a high quality, and related to the model you’re trying to build. For example, if you want to build a model that predicts weather by the hour, your dataset must include hourly weather data.
AI developers like us will work with you to ensure that your data is properly configured to make the most of your AI investment. If you’d like to spot-check whether your data is any good, check out Google’s Data Preparation Checklist.
Staff Buy-in
Get Leadership Excited
Let us state emphatically: successful AI initiatives require enthusiasm from leadership. In fact, we’ve found that one of the highest predictors of AI success is whether someone in the c-suite is playing the role of AI evangelist.
See, AI adoption is not a short-term play. It requires foresight, strategic thinking, and an investment of thousands, sometimes hundreds of thousands of dollars — and if you don’t have leadership onboard, it will be difficult to get the institutional support to meaningfully bring about AI transformation.
If you are spearheading your AI deployments, we’d like to point you to our AI resources page. Each of the ebooks and articles that we’ve gathered there have been carefully designed to make it easier for you to make the case to your leadership.
Getting buy-in from leadership can take a while. Most leaders intuitively understand AI’s usefulness but may need additional education when it comes to realizing AI’s full potential. This is another reason we recommend starting with an AI pilot project. By launching a successful proof of concept, you’ll have a stronger claim to rolling out AI more broadly.
Get Workers
Excited
For several years, we’ve been hearing about the possibility that AI could lead to massive layoffs. Self-driving cars, for example, could be the end of the trucking industry as we know it. Thus, when employees start to hear rumors about an AI initiative, they can start to worry that layoffs will soon follow.
Fortunately, recent reports suggest a different reality. For example, a Dun & Bradstreet survey found that only 8% of companies were cutting jobs because of artificial intelligence. And while many AI-related economic surveys predict that there will be job disruptions, disruptions don’t necessarily mean that jobs are going away.
In our experience, another important aspect of AI deployment is ensuring that your workers are not only prepared for the change but also excited about it.
What’s great about AI is that it makes boring and repetitive tasks basically effortless — and this ends up being absolutely huge for workers whose days are full of repetitive, often menial tasks.
Our clients, Allmed, had a similar experience. Before AI, an army of licensed physicians had to manually pour through patient health records to make treatment decisions.
Now, with natural language processing, these doctors are able to focus on the thing they do best: evaluating doctor-prescribed treatment plans and ensuring their patients are getting the very best care. AI freed Allmed physicians to use their time elsewhere.
To be sure, bringing AI to your organization will require your teams to adapt. You can expect the deployment of AI to bring with it new tools that your team will need to know how to use.
When securing buy-in from your workforce, be sure to address their concerns early and communicate often about the transition process. Eventually, you’ll want to be sure they’re armed with the proper training but for now, the goal should just be to get them excited.
Chapter 6.
How does the AI development process work?
When embarking on an AI initiative, many companies are keen to understand how the development process works. In many ways, working with us is similar to working with any software development organization. First, we’ll help you define the project and its goals, research correct solutions, build and test your model, and ultimately put it into production.
Step 1: Define Project Goals
As we’ve discussed in previous chapters, artificial intelligence is great for automating repetitive tasks, scaling the work of humans, and transforming datasets into predictive and actionable insights. If you’ve gotten this far in our guide, it’s possible that you already have an idea for an AI solution you’d like to build.
For those that don’t, we recommend you start small. When thinking about your first AI deployment, try to find a solution that uses data you already collect and is directly related to your bottom line.
In fact, during the discovery process, our investigations typically orbit around two main questions:
- What are your long-term strategic goals?
- What data do you have?
Thinking holistically about how your company operates and where it’s headed can give you some excellent insight into what kinds of cognitive intelligence solutions would be good for your organization.
Furthermore, by paying attention to the kinds of data you collect, you should be able to identify a few opportunities for quick AI adoption.
For example, a manufacturer might build an AI solution that detects manufacturing defects using the video cameras already installed on the factory floor. This would be a great AI pilot project because it uses the data they already have (video camera footage) and is directly related to their bottom line.
Having existing data streamlines the AI adoption process. However, if you don’t have the data you need, we can help you develop and execute a data creation plan.
And if you’re completely unsure where to place your AI bets, we suggest you check out our free resource, Discussion Questions for AI Readiness. There, you’ll be able to swim around in dozens of thought-provoking questions that will help you imagine the ways that AI could make a huge difference at your organization. The questions are organized around solving specific problems related to:
- Planning and decision making
- Operational optimization
- Customer service
We focused on those three as lots of research has indicated that those are the key value drivers for AI adoption.
Ultimately, making decisions around AI can be tough because the technology is changing so quickly. Here at Manceps, we love helping companies through this process. Our AI Applied service combines enterprise strategy with AI technology engineering to help your organization develop and execute an AI transformation strategy to maximize your ROI.
Step 2: Conduct Data Audit
Once you have defined the scope of your project, our AI engineers will go to work determining your AI project readiness by looking at your datasets and ensuring that they are properly structured and well-aligned with your AI goals.
See, artificial intelligence depends on having the right kind of data in the right format. For example, if you’re looking to make hourly weather predictions, then your dataset must include hour-by-hour weather information.
Step 3. Research AI Solutions, Algorithms, and Existing Models
Once we determine the shape and quality of data, we then embark on a discovery process to stitch together an AI solution. Fortunately, the space is populated with a vast set of pre-trained models and open-source technologies that make building an AI model easier than ever.
However, AI research is clipping along at a breathless pace. Models that were industry-standard a month ago can be outdated a few weeks later. The best AI teams will integrate these cutting-edge capabilities into the solution they outline.
Step 4: Package the Project for Stakeholders
This is the stage at which an entire AI deployment plan comes together. Here is where all of the project requirements are gathered, the models selected, the Machine Learn infrastructure agreed to, and everything is mapped out. It’s also the stage where you’ll want to make the business case for stakeholders and others involved in the process.
Once everyone agrees that this is the right direction, the next step is to begin building out the model.
Step 5: Design the Model
Here is the part of the process where we train and optimize the model. Because AI models depend on data, you can expect a big part of this process to include data structuring and labeling.
Once properly organized, the data can then be fed into the model for training purposes. We should mention at this point that training a model can be a resource-intensive and expensive process. We use a variety of on-prem and cloud solutions to ensure that your model gets trained as quickly and affordably as possible.
At this point, the model will be functioning at a baseline level of accuracy — usually around 70%. From there, depending on your particular use case, our team will work to refine the model further and further until it functions as intended. This testing process can take as little as a few days to several months, depending on the complexity of the model you’re trying to build.
Step 6: Operationalize the Solution
Once the model has been trained and optimized, it’s time to integrate it with your business operations. In most cases, this involves creating the proper APIs so that your model can communicate with other systems. These systems can then perform tasks automatically or relay information to human-users.
Chapter 7.
What will an AI shift look like?
For most of this guide, we’ve explored the ways that artificial intelligence can transform your business operationally — but one thing we haven’t quite discussed is how AI can change things culturally. The transition to computers in the 80s and 90s is a representative example of how new technology can completely change the way companies are organized, what skills workers need, and how leaders make decisions.
In this chapter, we’re going to preview some of the cultural transformations that AI will bring to your organization and how to prepare for them.
Three Major Cultural Shifts
Before diving into how to manage the shift to AI, we wanted to preview the major shifts that you can expect your company to undergo — namely, that AI will shift how teams will work together, how decisions are made, are your company’s tolerance for risk. Let’s explore each in more detail below.
AI Will Change
How Teams Work Together
AI sits at the intersection of data analytics and operations, which is to say that bringing the technology to your organization means that cross-functional teams are going to have to find ways to work more closely together.
For example, a quality assurance solution would need to gather project requirements from the on-site team, the analytics team, and the product teams.
This cross-departmental collaboration not only makes for better AI solutions but also for better organizations.
AI Will Change
How Leaders Make Decisions
Artificial Intelligence will shift decision-making from those with the most experience or the best instincts to those with the most data. By bringing more automation and data to the decision-making process, companies can expect a shift in how organizations chart a path forward. Pre-AI, we can expect the majority of decisions to be made via traditional top-down structures. Post-AI, we can expect companies to locate decision-making at lower levels.
We are already seeing this with the rise of AI-powered stock trading. Companies are already starting to favor the insights of quants over their most profitable and experienced traders.
AI Will Increase
Risk-Tolerance
As we’ve previously mentioned, because artificial intelligence becomes increasingly capable over time, we can think of it as an appreciating asset. This creates urgency to move quickly in order to capitalize on the benefits of any given model. Thus, we can expect that companies that have an AI-facing culture will take the lesson of innovating quickly to heart.
How to Ensure A Smooth Transition to Artificial Intelligence
In Chapter 5, we discussed the importance of securing buy-in from both leadership and your staff before embarking on an AI initiative. This effort, while commendable, is just the beginning. At Manceps, we’ve found that different clients treat this part of the process differently to mixed results. The companies who see the most benefits from their AI deployments are also the ones who took this part of the process seriously.
Educate Your Team About AI’s
Possibilities and Misconceptions
Over time, we’ve come to realize that the companies that experienced the smoothest AI transition were the very same that invested in AI education throughout the process.
Resources like this can help your employees wrap their head around AI and start to imagine possible AI initiatives they may wish to deploy in their own departments.
Any educational component should not only focus on the impact AI will have on your organization but also on the employees themselves. When companies bring forth greater AI-powered automation, workers can expect to spend less time on repetitive tasks, freeing them up for more valuable pursuits.
When planning to educate your organization about AI, we offer these tips:
- Tailor the message for different stakeholders throughout the business but don’t limit your AI education initiatives to those directly involved in the project.
- Focus on the why as opposed to features of the technology.
- To correct misconceptions, be sure to spend a little time differentiating between AI’s (more boring) corporate applications from its R2-D2/Terminator pop culture associations.
Create a Transparent
Map for AI Deployment
Avoid problems along the way by creating a clear AI adoption plan. By being transparent about the roadmap for the project, employees can get used to the transition over time. This allows for a step-by-step adoption as opposed to an all-at-once approach.
In the beginning, you can roll out your new AI solution to tech-savvy stakeholders or teams. By offering the solution first to early adopters, you are effectively creating an army of evangelists that can then encourage your wider teams to try the new technology with an open mind.
Remember, AI requires internal champions — so try to locate those champions among those who are particularly interested in new technologies. Bring them into testing initiatives to give leaders a glimpse into how everyone else might react.
By going slow, you give your employees time to adjust to what may otherwise feel like an abrupt and radical change.
Treat Education as
Ongoing and Iterative
Change management doesn’t end when your AI solution launches, which is why we always advise our clients to engage in ongoing education to maximize their investment in AI.
To do this, try to push beyond the usability standard to the value standard. What we mean is that the goal of ongoing education isn’t just to ensure your team knows how to use the solution but also that they see the value in it and how it helps them.
Uncertainty can cause a lot of issues for your organization. If you invest heavily in deploying an AI system but then fail to train your people, you will be leaving many of AI’s benefits on the table.
There are many different ways to train your workforce. We’ve seen it done via documentation, tutorials, webinars, and youtube videos. The trick is to find something that works for your organization and stick to it so everyone can enjoy the benefits of AI.
Interested bringing AI to the enterprise?
Get exclusive AI guides and resources.
Need Some Ideas?
Check out our AI examples series.
