Newsletter #21: AI can be easy to say but hard to define. 7 Q&A's to make it real.

Artificial Intelligence is an easy term to say but an increasingly difficult term to define. Especially in practical terms that resonate with senior non-technical business leaders.

Day-to-day, I often see senior leaders on either end of the tech quotient spectrum inadvertently talk past each other, which is understandable given the pace of exponential change we’re living through.?

But like technical debt, the challenges it creates compound over time which can create real problems down the road.

Newsletter #21 is going to be a bit different from others because I’m going to try and make AI more tangible by walking through a series of Q&A focused on Intelligent Systems.

A topic that’s foundational to all of the hype / noise / signal around Generative AI.

I hope this might help readers that sit on either side of the senior non-technical or senior-technical find more common ground.?

And for those that are already good to go, good for you and hopefully there’s a nugget or two in here that might enhance what you all are already doing.

1: What is Artificial Intelligence?

AI refers to the simulation of human intelligence in technology systems that allow them to take actions that otherwise would be done by humans.

The systems can be designed to automate tasks and automate aspects of human workflows, creating substantial business benefits.

An AI system’s architecture typically includes multiple components that work together to process data, learn from it, and make intelligent decisions (i.e. Next Best Action).

2: What are the main components of an AI system?

This list isn’t exhaustive but core AI systems include the following 8 components / categories:

1: Data Collection + Preprocessing

Gathering data from various sources, data cleansing to ensure accuracy and data transformation into a set of attributes that can be understood and processed by AI algorithms.

2: Algorithms + Models

Machine Learning algorithms are mathematical models and techniques used to train the AI and validate its accuracy.

3: Computing Infrastructure

Hardware such as Graphics Processing Units (GPUs) can accelerate AI computations and Cloud Platforms are leveraged by AI systems for scalability and storage.

4: Deep Learning + Neural Networks

Deep Learning is a subset of Machine Learning that leverages a Neural Network architecture to process and learn from vast amounts of data which enables an AI system to perform tasks that used to be exclusive to the human brain.

5: Natural Language Processing (NLP)

A branch of AI, NLP is focused on enabling humans and computers to interact using natural language so computers can understand, interpret and generate human language that is accurate, meaningful and useful.

6: User Experience (Interface + Interaction)

User Experience is the space where humans and AI systems interact and includes the layout, design and visual elements that users interact with while also providing feedback and the associated value / utility generated for the user.

7: Feedback + Continuous Learning

The way an AI system gathers feedback, drives continuous learning and optimization based on new data being gathered and understood.

8: Ethics + Bias

Ensuring that the workings of the AI system can be understood and explained while biases can be identified and corrected.

3: What is Generative AI?

Generative AI (LLMs) is a subset of Deep Learning + Neural Networks and enhances that component of an AI system by producing new, original content based on the patterns of the original input data the models and algorithms were trained on.

The models identify patterns, structures, and relationships within the training data so they can produce new data that wasn’t in the training set but resembles it.

Business applications include content creation, simulation and personalization of experiences, designs and / or tailored product recommendations.

The transformative potential of LLMs to evolve the computing landscape is driven by their ability to understand context, generate content and eventually train themselves which represents a significant leap in AI system capabilities.

4: How could Generative AI enabled AI systems impact how companies think about their tech stack?

Don’t make the mistake of thinking about Generative AI / LLMs as chatbots as that’s the equivalent of thinking about computers back in the day as fancy, high powered calculators (s/o Andrej Karpathy )...not sure why that one resonates so much with me but it makes me laugh every time! ;)

Oversimplifying it but consider your tech stack as the foundation upon which your business is built, supporting the various departments, teams, functions and people who have goals / outcomes they’re trying to generate.?

Each of the goals / outcomes include various tasks that when combined can be referred to as a workflow where humans leverage different types of technology to get things done. Increasingly, those technology parts are integrated into AI systems that support various workflows and improve via feedback loops aka the more you use them the better / more useful they get.

Contextualize those goals / outcomes and workflows by thinking about the way things got done before 11.30.22 (ChatGPT launch) and after 11.30.22, assuming your company has started integrating Generative AI enabled systems into your tech stack.

The transformative power of Generative AI is driven by the fact that it’s a General Purpose Technology which means it stretches horizontally across an organization, creating all sorts of improvements and positive benefits across your business vs being applied more narrowly.

The devil is in the details of where / when / how / why you apply Generative AI enabled systems.?

The good news is all of your existing tech stack players got the memo and are moving as fast as possible to modernize their features / capabilities by embracing Generative AI in an attempt future proof their product by ensuring they’re a component part of the modern, Generative AI enabled tech stack that all companies will eventually embrace.?

Some sooner than others but it’s safe to say that it’s not an if but a when at this point.

5: Can you bring the above to life using Generative AI enabled Customer Experience?

In June, 麦肯锡 published “The Economic Potential of Generative AI” (link) that outlined “about 75 percent of the value that Generative AI use cases could deliver falls across four areas: Customer operations, marketing and sales, software engineering, and R&D.”

I broke down what I found to especially interesting from the McKinsey report in Newsletter #8: "McKinsey bringing a whole lot of signal in their 68-page report - 'The Economic Potential of Generative AI' --> 5 key takeaways" (link), but specific to this question, I think there are 3 areas of opportunity that deserve immediate focus:

1: Give Your People Superpowers

“Industrial-age machines enabled workers to accomplish physical tasks beyond the capabilities of their own bodies. More recently, computers have enabled knowledge workers to perform calculations that would have taken many years to do manually.”

2: Eliminate One Day Per Week of Tedious Work

“In 2012, the McKinsey Global Institute(MGI) estimated that knowledge workers spent about a fifth of their time, or one day each work week, searching for and gathering information.”

3: Recalibrate Your OpEx Assumptions

“We estimate that applying Generative AI to customer care functions could increase productivity at a value ranging from 30 to 45 percent of the current function costs.”

To capitalize on the 3 areas of opportunity, there are 3 drivers to consider immediately:

1: Knowledge Management

Leveraging knowledge bases and knowledge graphs to enable a more holistic, context-aware and personalized approach to customer experience including anticipating customer needs driven by predictive analytics. Creating previously unimaginably customized experiences delivered by a previously unimaginable OpEx number.

2: Self-Service

Generative AI transforms self-service platforms from static, one-size-fits-all solutions to dynamic, personalized, and interactive experiences. Empowering customers to find solutions efficiently while reducing the burden on customer service representatives while building towards the north star, a seamless omnichannel customer experience.

3: Pareto’s Principle

A framework for how to think about humans + machines collaborating to optimize for maximum impact on the 20% of inputs that drive 80%+ of the results.

Are your humans being enabled by your Generative AI Enabled Systems to do what they do best including building relationships, discovering deeper / nuanced customer needs, designing complex strategies, completing more empathetic / creative tasks etc? If not, why not?

6: To get tactical, what are the top 5 questions that a cross-functional team reporting into a business unit leader could use as a jumping off point to create a 90-day tactical plan to put points on the board?

1. Are there any manual or repetitive processes tied directly to revenue generation that you believe could be automated or optimized?
2. Which software or tools does your team use daily and which ones impact your top 3 most important priorities / initiatives?
3. What data sources does your team have access to, how is it processed into insights and can you provide 3-5 examples of how those insights are regularly applied to critical decisionmaking?
4. What feedback loops exist between the technology team, end users and business leaders around the role your technology stack is or isn’t playing in enhancing priority workflows that contribute directly to priority business outcomes?
5. What is the process for evaluating emerging technology that can be introduced to make your AI system more of an Intelligent System that generates more of your priority business outcomes?

7: Last but certainly not least, how does an individual executive get their Generative AI enabled Intelligent Systems quotient way up?

Sure, you can read about it but if you wanted to learn how to swim or ride a bike, what would you do? This stuff is no different.?

Quote from the blog of one of the leading thinkers on all things AI, Ethan Mollick - “What people ask me most. Also, some answers.” (link) - hits the nail on the head.

“But, generally, my recommendation is to follow a simple two-step plan. First, get access to the most advanced and largest Large Language Model you can get your hands on.”

“Then use it to do everything that you are legally and ethically allowed to use it for. Generating ideas? Ask the AI for suggestions. In a meeting? Record the transcript and ask the AI to summarize action items. Writing an email? Work on drafting it with AI help.

My rule of thumb is you need about 10 hours of AI use time to understand whether and how it might help you.

You need to learn the shape of the Jagged Frontier in your industry or job, and there is no instruction manual, so just use it and learn."

"I do this all the time when new tools come out. For example, I just got access to DALL-E3, the latest image creation tool for OpenAI. It works very differently than other previous AI image tools because you tell ChatGPT-4 what you want, and the AI decides what to create. I fed it this entire article and asked it to create illustrations that would be good cover art.

"And here is what it came up with:"

“For the things we have to learn before we can do them, we learn by doing them.”

- Aristotle

That’s it for this week. Hope you found it useful and have a great week!

Talk soon.

Alec