FinText training sessions on using AI in daily workflows. Clients include AWS and Aviva Investors.
Large Language Models, now colloquially referred to as AI, seem to have taken the world by storm. But when it comes to day-to-day business operations, many companies are still wondering:
Where, exactly, are those magical productivity gains? And how does one go about capturing them?
FinText has been educating clients from the early days of publicly available AI tools on how to put them to use – especially for creating B2B content.
Each company has its own processes and workflows. Everyone uses Excel, for example, but the specific spreadsheets differ from company to company. Companies can therefore expect that where they value in AI may not match what others happen to be doing.
Equally, different teams have different appetites for new tech. Some teams are keen to discover and test new tools, others want to get a more solid footing on what all this noise is about.
For these reasons, FinText’s training on putting AI to work in business always starts from understanding where the clients are at.
For example, one training session was hosted by AWS to representatives from several different companies. While the attendees had technical awareness of what large language models are, their interest was in communicating to their own companies what could be achieved with them in practice.
A different training session, delivered to Aviva Investors, focused on first introducing the fundamentals of how LLMs operate – knowledge that is useful to get the most out of any of the AI tools out there.
We were asked to design our training sessions with this goal in mind: give participants practical skills in choosing where to apply AI solutions, and how to design a process to make the best use out of them.
To do this, we always begin by establishing and understanding what this thing, this “AI”, has actually learned to do.
It would seem like it’s learned perfect English. But if that were the case, why does it always do certain things, like hallucinate or phrase sentences in a tinny, robotic tone of voice?
We then follow by thinking about how to structure tasks around what an AI is capable of achieving. We don’t just stick to abstract ideas, but offer practical, step-by-step examples of doing this.
From there, we move on to naming and playing with some of the most useful tools available.
We always recommend to clients that the training include active experimentation, because fear is a substantial and rarely discussed barrier. We find that joint experimentation helps alleviate concerns about “doing it right”.
The training was very interactive, with live demos of how to use the GPT-3 model to enhance content creation. Vered covered how these models ‘think’ and how their abilities are best applied to business-writing scenarios.
I enjoyed the discussion on the practicalities of generating synthetic content. This approach can work well for professional-services companies looking to scale content creation.Heiko Hotz, Senior Solutions Architect for AI & Machine Learning, AWS
To learn more about using AI tools in your everyday work – please contact us here.