Accessible AI Assessment
Accessible AI is a way to reach your strategic goals faster and cheaper than ever before. But it is not without challenges for leadership, including for many a new way of imagining the speed and potential for progress. Adopting AI means embracing data-driven decision making, having the right technology stack and deciding whether to build or buy. Let’s look at each in turn.
Model capability is scaling rapidly. Reasoning models that perform at the highest levels of human performance in coding and academic tests are now available at low cost. Yet there is still no transparency about how models work and why particular outcomes arise. You really must trust your data. For this reason, it is essential to work with diverse and high-quality datasets.
Thereafter, your business must be able to handle AI. Technology may be owned or leased, and MSBC can advise you on your options and set it up for you. You will need a means of communicating with your data, which might be a chatbot, existing software services such as Tableau or PowerBI, or generative AI text, image, audio and video outputs in social media. We have a range of test cases for you to try out.
Only technology giants and well-funded governments have the means to build their own large language models. Even adapting one of them requires data scientists, developers and continuing support. These resources may be outsourced, but reengineering a business is a continual process. Choosing to work with a partner with access to expertise, compute and talent may be your best option.
The alternative is to wait for the software tools you buy to adopt AI. This may allow tasks to be done faster but ties you into someone else’s data and design. It is also an open question whether AI communicating directly with databases will replace software, including spreadsheets, dashboards and CRMs. Our experience is that AI delivers meaningful cost savings when individual models are combined in continuous processes.
For instance, you may buy software to summarise meetings and write minutes. This saves time and are there several tools on the market that do this well. But do they save money? Until the model connects to other models that automate the priority tasks from the meeting, there is often no cost to take out of the business. This is why strategic planning is the first step in our consultations with clients.
You may view AI as if it were a new employee. They are keen and intelligent but need direction and oversight. They learn quickly and will soon work on basic tasks without supervision. At this point they are ready to take on more challenging roles.
We recommend starting out with a high impact project that has clear goals and a means of measuring them. We are happy to work with you to determine what this will be and offer a free consultation to help you decide.