AI Policy & Ethics

Connected understand and actively embrace the opportunities that AI can offer in terms of labour-saving, insight, data-processing, economic growth, nature of work, and invention. We are also very aware of the risks it poses.

We are not the builders of AI substrates. We evolve constructs on the substrate – specifically LLMs – to enhance or refine their function in-line with organisational needs.

We are fully committed to the adoption of non-prejudicial systems – specifically around computer vision and natural language processing – and recognise the challenge of the unfettered use of unsupervised Reinforcement Learning.

With a “do no harm” stance when it comes evolving AI models, we leverage Machine Learning as part of our services offerings.

This includes areas such as traffic management, predictive modelling and more in order to transform service delivery for clients. Backed by valuable insights from a decade worth of support tickets there’s an abundance of data available which we use to enhance the client experience.

Our marketing services utilise Frontier AI technology for maximum efficiency and accurate results. We are committed to providing top quality service by harnessing the potential of artificial intelligence in content creation/optimisation, media production, campaign administration, SEO, testing analysis and reporting. As an organisation dedicated to growth through innovation we remain open to adopting further advancements in this field that would elevate our clients’ experience even more.

Bespoke Trained GPTs

It is important to understand that we use brand and subject specific trained models based on OpenAI – we do not use generic versions.

For example, for our own business we have organically grown a bespoke model trained on 1000s of data points including every piece of content we’ve published for the last decade, our tone of voice, brand values, and core skills.

This creates a form of highly-contextualised super-intelligence that our teams can use for all manner of tasks.

We also do the same for most clients. These models can be used to increase internal efficiencies, other times it is used to directly assist client teams and functions.

If you are serious about adopting AI in the digital sales/marketing functions at your business then please do contact us to find out more.

Typical bespoke models we’ve built for clients use 1k datapoints which is, in our opinion, the tipping point where bespoke trained models make a dramatic difference to service deliverability. Some models we’ve built reference millions of data points.

AI systems learn from their training data, which can unintentionally include biases. While many language models use filters to mitigate bias, these are not foolproof. We must review and develop content to ensure it is inclusive and accessible. Therefore, our company policy is that we should NEVER publish or send something that has been written entirely by AI without human development or review for quality and accuracy. Furthermore, AI should not mislead or manipulate customers; generated content must align with corporate values and undergo review for bias.

We understand that little or no regulation exists in the AI space and therefore follows a number of governance leaders in the field including The Ada Lovelace Institute and the UK’s AI Safety Institute.

This entire page was originally created in February 2020 and subsequently updated fairly furiously ever since!

Page updated: 2024-06-13