How telcos can put the “i” (implementation) into AI: part 1

It’s almost two years since OpenAI announced the launch of ChatGPT, starting a transformational gold rush that has seen companies of all shapes and sizes scramble to get a handle on what AI can do for them. But it’s a race where the winning post seems to remain tantalizingly out of reach as the big names in AI development keep on pushing for the next big breakthrough. As one example, OpenAI revealed plans last week to release its next flagship model – teased as potentially up to 100 times more powerful than GPT-4 – by this December.

The speed at which things are changing is impressive, but for the average business on the ground, have even the fledgling versions of these new technologies proved worthy of their potential? Some studies suggest not: according to KPMG , only 15% of organizations have metrics to track their return on investment in generative AI, while less than half (44%) have progressed beyond the experimentation stage to scale up their use of the technology.?

It's an issue I have seen many times before, with many different tech advances. And the problem isn’t to do with the technology itself. It’s actually about getting the implementation right, which means gaining enough short-term value to gauge long-term potential.

This is where a lot of companies fall short – but in my experience, it doesn’t have to be that way. To implement AI successfully, leaders should keep six “golden rules” in mind. While mostly based on what has worked for telco companies, there are parallels that can be applied across other types of organization. In this article, let’s take a look at the first four rules, which are all about getting organized internally.

Rule #1. Don’t assume you need a massive budget

From what I have observed to date, the beauty of AI implementation is that it does not need to be very expensive, as direct tools such as ChatGPT and those offered as an add-on to existing software (such as Copilot for Windows) have demonstrated. It is also a technology that can be used to offer benefits in many different areas, meaning you can experiment and move quickly without a lot of money, resources or complexity.

Rule #2. Use your own corporate superhero to lead the charge

While there are obviously different execution options, I would recommend starting by empowering your CIO or head of Digital to lead on developing the organization’s initial AI plans. With security risks, the choice of applications, and the necessary connection with your own databases, this will require a lot of IT involvement. My next tip is for that person to establish a small team of cross-functional AI “super users”, backed by dedicated IT support. We all know those early adopters who are fascinated by advances in technology and have probably jumped on the bandwagon already. Find where they are in your most critical functions and get them on board.

Rule #3. Get the right governance and key principles in place

Once your AI “A team” is in place, they need to establish critical ground rules like how decisions will be made and how progress will be reported both to the leadership team and across the organization. Other key principles should include:

• How internal data will be treated (i.e. how it will be protected from flowing into external models that could create commercial risk);
• How projects will be prioritized and approved (i.e. what the requirements around business goals, objectives and approvals will be);
• How funding will be agreed and allocated;
• How specific AI tools will be selected for implementation.

Rule #4. Announce the company’s plans and how they will be managed

Now, you are just about ready to get motoring. But first, it’s important to communicate how you are going to go about testing, implementing and scaling AI across the company, who is leading the effort, where the first projects will be deployed, and why. This should also make it clear to all employees where they fit into the AI equation, what is expected of them in terms of responsible and ethical use of the technology, and how any inappropriate behavior (such as misuse of data, or out-of-scope side projects) will be dealt with. The core team could also develop some training and guidance for the wider organization.

With these foundational tasks complete, the next big questions are where to implement first, what will help you get it right, and which risks to guard against. We’ll delve into those next time.

Until then, enjoy this article about what happens when AI gets into podcasting . Could the game finally be up for, well, everyone else out there in the podcast-sphere? Time will tell…

?

Photo credit: The KonG, Shutterstock