As a CEO, I’ve seen so many allegedly “game-changing’” technologies come and go this century – from AR/VR technology, blockchain, 3D printing, the Metaverse, NFTs, Web3, and the list goes on and on. Now, Generative AI is at the forefront of every business owner’s attention.
Just like in the past, many companies are rushing to invest substantial sums just to say they are focusing on Gen AI, all without even defining a desired business outcome.
On the other end of the spectrum, those companies are living in fear that they are falling behind but are overwhelmed with the possibilities of AI and don’t know where to even start.
These companies recognize they need AI, but they’re similarly being distracted by the narrow band of Generative AI. In my experience as a CEO, building a small two-man shop into a multi-national, multibillion-dollar enterprise – for a new technology, such as AI, to pay; there needs to be a clear business outcome and a clear target for adoption.
AI is here, and it will grow, but AI is also a very different technology with different business outcomes for each company. Rather than speeding to a pilot, we recommend focusing on hustling to a plan. Full AI Discovery allows your organization to properly plan where AI makes sense, look at ALL AI tech available, and make investments with confidence and purpose.
We’re seeing three common themes during Discovery: first, data is often a roadblock to most implementations; second, Generative AI can only solve a small number of problems business owners highlight; and third, it’s important to objectively decide which AI pilot launches first.
Deep Learning and Machine Learning Are Solving More Problems Than Generative
We’ve interviewed hundreds of companies that have come to us thinking they need generative AI to solve all of their problems. That only covers a small portion of the issues these businesses need help with, and it’s honestly a distraction from solving their biggest concerns.
One common theme we hear in Discovery is businesses want to be able to crunch numbers, identify patterns, and make predictions based on the data they have. If a business wants to solve for those three things, their best bet will be using deep learning, machine learning, or a combination of the two – not generative AI.
Machine learning is still a much more prevalent and relevant technology for analytics. Businesses- where humans have identified a hierarchy of features in their data- are already ready to leverage machine learning. The best example is how human-defined product categories and features allow Amazon to recommend the next product to buyers. Likewise, machine learning is the right AI for things such as anomaly detection (think cybersecurity or fraud detection), curating content for users, dynamic pricing, or even medical diagnosis.
If a business has a lot of unstructured data and wants to make sense of all of it, deep learning is the right AI tool. Unstructured data includes things as simple as a text document. Because there can be infinite variations of a single document, you need an AI that can comprehend the data and make general observations on it without manual intervention.
For instance, many businesses collect files from their clients and then have to manually input the data into their system. Imagine a CPA that receives a file of uncategorized receipts from a client. Deep learning can comprehend the data and extract the appropriate information to input it into the system without human intervention. Deep learning is what puts intelligence into intelligent automation.
Data Can Make or Break Your AI Implementation
Right now, when people talk about using AI in their business, more often than not, they’re using a ChatGPT like LLM to answer questions. The problem is publicly available LLMs are not trained on the specific data of individual businesses. To have an AI that is truly valuable and not just a novelty, businesses need to be clear on the data they want to leverage.
Many businesses have started down a path of cleaning and storing their data – from data warehouses to data lakes to data marts. These data projects are often necessary to have usable data for custom AI models. But, for every business that has completed this process, there are probably 1000 more that haven’t. Now, generative AI threatens to distract from finishing out those projects. Worse, a lot of people are under the impression that generative AI is a solution to their data problems.
Look no further than Google’s recent misstep with their own generative AI, Gemini. The search giant was forced to roll back their ChatGPT competitor after several humiliating blunders went viral, including a recommendation for the minimum number of rocks a person should eat a day as well as a charming glue-based recipe that keeps cheese from falling off pizza. The problem? Google trained Gemini on all data across all Google searches, including joke social posts.
The simple lesson is that you can’t just throw all your data at AI and expect everything to work. You need to be deliberate in choosing your data, and you need to make sure your data is clean. Again, AI Discovery can help businesses figure out what they can do with the data they have and how they can get more out of their data.
Objectively Deciding on Which AI Projects to Start
You may find that your organization has tens or even hundreds of potential use cases for AI – some of those use cases may even be Generative AI-based. How do you decide which ones to pilot?
We often find that the loudest voice in an organization ends up deciding which AI project gets immediate attention, but that project is often not the best option when every idea is weighed in Discovery.
During Discovery, businesses should be collecting every idea and use case for AI across the organization. Then, based on the goals of the organization, each use case should be objectively scored to determine priorities.
The Importance of AI Discovery
Imagine you’ve just heard about a new chain of tropical islands that everyone says is the most incredible vacation experience, and you set out to go. Except, you have no plan, no budget, no mode of transportation, and you don’t even know where they are.
That’s what businesses are doing with AI. Diving into AI projects without a structured discovery phase can lead to misaligned priorities, wasted resources, and missed opportunities. AI Discovery provides that essential “map,” guiding businesses through the complexities of AI implementation. Three of the essential outcomes of AI Discovery are:
- Inventory of current technological landscape – Businesses should identify their existing capabilities as well as gaps that need to be addressed.
- Demystify the selection of the right AI for each problem – Gaining a deep understanding of different AI technologies—such as generative AI, intelligent automation, and deep learning—businesses can make informed decisions.
- Prioritization of AI projects based on strategic business goals – Rather than chasing every possible AI application, Discovery allows businesses to focus on initiatives that promise the greatest return on investment by assigning objective scores to each use case.
The outcome of Discovery is a clear roadmap for implementation. It’s not enough to just get a report of where AI would be helpful, you also need a strategy for actually deploying it and measuring its impact. This process needs to lead you to a cognizable benefit to your business, if it’s not, it’s just hype like all the other technologies that came before it.
