Our team members are now certified prompt engineers for AI – what does that mean?
As you may be starting to understand by now, at Torque we’re taking AI seriously. We believe it can be a force for good for the communications industry and we’re proactively employing AI in the work we do for our clients and as an agency (read more on that here).
.
While the potential of AI is limitless for communications, it must be used appropriately, responsibly and knowledgeably. That’s why four members of our team are now certified prompt engineers, with all team members set to achieve certification in the coming months.
.
Following an intensive training period, which saw team members learn, develop and test a full range of ‘prompt patterns’ for use in AI, they received certification from one of the top 15 universities in the USA. The certification enables Torque’s team to use AI effectively, ensuring they use proper processes and follow due diligence.
.
What is prompt engineering?
Anyone who has used a Large Language Model (LLM), such as ChatGPT or Google Bard, will know that it often feels like you are having a conversation with it. Looking at it like this, it’s easy to be fooled in to thinking you can ask an LLM anything and, by some wizardry, the output you get from it will be perfect. That is not the case.
.
While an LLM like ChatGPT is good at mimicking conversation, everything you ‘say’ to it is a command, or prompt. LLMs have been designed in such a way that they respond to some prompts better than others, depending on the use case and the type of output required.
.
This means that the best way of achieving a high quality, accurate output from an LLM is by designing a high quality, accurate input. There are various prompt patterns that can be employed when prompting an LLM, which allow the tool to properly understand and process the task you need it to complete.
.
By using AI tools in the way they were intended – and by tapping in to how they ‘think’ and work – we can improve the reliability and accuracy of the content or output we get from them.
.
What else are we doing?
We’re fully invested in ensuring we use AI responsibly, appropriately and transparently.
.
As well as training our team to become certified prompt engineers, we are upskilling several team members with training on specific AI tools. For example, some members of our team have also completed training that helps them get the most from image generation tools such as Midjourney.
.
Our team also comes together on a regular basis to share ideas, develop new use cases and, importantly, troubleshoot any issues they may have. To ensure we continue harnessing our team’s growing expertise, we also manage an agency-wide prompt library, which catalogues and ranks the effectiveness of all prompts generated by the team.
.
This work is all underpinned by our Five Principles for AI in Communications, which we have developed to ensure we always use AI ethically, safely, and accurately.