CTO insights with James Akrigg

Dec 7, 2023 | Guest blog, Technology

The rapid evolution of Artificial Intelligence (AI) is completely reshaping the business landscape, presenting both opportunities and challenges.

At our recent Harold Sharp staff Away Day, we invited James Akrigg, a seasoned virtual Chief Technology Officer (and self-confessed Chief Technology Optimist) to share his experience and insights, drawing on 30+ years in the technology sector. Focused on ‘leading with AI – now and in the future’, James condensed 70 years of R&D into a thought provoking 30-minute presentation.

Much of the content from his session is relevant to SMEs as they navigate the AI landscape. In today’s blog, we distil these insights into four key AI lessons for business owners.

Lesson 1: Validate AI responses for informed decision-making

In the business realm, relying on generated AI responses without validation can lead to costly errors.

Using examples like OpenAI ChatGPT, Google Bard and Microsoft CoPilot (Bing Chat), James emphasises the need for iterative interactions with AI systems. This is because AI can suffer with ‘hallucinations’ (which effectively means that it can go off course and start making things up).

He also reminds us that the machine does not think for itself, it uses the syntactical and the semantic structure of language to predict the next word in a sequence. He urges business owners to treat AI as a collaborative partner, refining its responses to ensure accuracy. This caution is particularly vital in professional settings where misinformation can have significant consequences for decision-making.

Lesson 2: Guide AI’s learning process for business-relevant outcomes

Linked to the above, despite AI’s advancements business owners play a crucial role in shaping AI systems.

James highlights that AI, especially generative models like GPT (Generative Pre-trained Transformers), rely on human input and validation. Business professionals should actively engage with AI tools, providing guidance to enhance the quality of outcomes.

Recognising AI as a tool to be refined by human expertise reinforces the importance of an active, hands-on approach to ensure AI aligns with business objectives. He stressed that the more that you, as the ‘trainer’, can input into the context of the conversation with the AI chat tools – the more accurate and relevant the output will be. Much like a search engine, the more precise the terms used in a search the more accurate the search results will be.  Creating the right prompt or prompting, the term used when requesting information from AI chat tools, needs to be more detailed to ensure the AI can respond correctly.

In an example, James requested an image of a car driving through Manchester. The prompt to produce the photo was more than 100 words. And, once produced by the AI tool, the output was four separate images of the car at different angles. By selecting the preferred image, James was reaffirming his choice and training the AI to recognise the human preferred outcome.  This is often referred to as RLHF (Reinforcement Learning from Human Feedback) and is how most generative AI systems improve quickly to meet human expectations.

Lesson 3: Seize AI as a strategic tool for service enhancement

James illustrates how AI presents a significant opportunity for businesses to elevate their services. By showcasing a beta demo using a custom GPT, something recently made possible in ChatGPT Pro, for streamlining expense management, he emphasises the transformative potential of integrating AI into business processes.

In his example, James built a fake expense programme. He then created a custom GPT to work with expenses and fed it his ‘company’ expense policy. The result? He captured a receipt and uploaded it to his custom GPT. The GPT vision read the receipt, analysed it, checked it against the policy document, published it correctly into the expense system and then provided a short summary of the actions it had undertaken. In a similar exercise, he fed AI the Harold Sharp website URL and requested the coding to build a website using the same structure.

The takeaway? You can now use AI to help compose your own software, rather than write it from scratch i.e. you don’t have to be a software developer to create custom GPT’s, but you do have to think like one! James encourages business owners to view AI not just as a technological advancement but as a strategic tool. He stresses that evolving skills and collaborating with AI tools can lead to the development of more sophisticated and efficient services, giving businesses a competitive edge.

Lesson 4: Navigate AI innovation with ethical considerations

As business owners explore AI applications, James stresses the importance of balancing innovation with ethical considerations. Recognising the potential and limitations of AI is crucial to responsible use.

Whilst leveraging AI capabilities, business owners must consider the values, truth, intent and bias that are aligned with the use of AI.  This is true for both the technology companies creating the AI models and the businesses implementing tools that build on these models.

Use of AI also raises challenges around consent and copyright as much of the content collated by AI is being scraped off the internet. Plus, there’s the wider impact of the compute power required and resulting carbon footprint that must be considered.  Understanding the ethical implications of AI ensures that businesses contribute to positive societal impacts while embracing technological innovation.

The conclusion?

AI is here today and it’s not going anywhere, so it’s important that everyone experiences what it can do as well as its limitations. So, if you’re a business leader reading this wondering where to start with ChatGPT, perhaps you might consider asking it to tell you a tax-related joke. And if it’s not funny, ask it to tell you another. And then, if it is funny, ask it to write a short social media post so that you can share the joke.

Jokes aside… business leaders should consider how they can leverage AI effectively, making informed decisions, enhancing services and contributing to both technological advancement whilst also maintaining ethical business practices. James’s 4 key lessons provide a good start point as part of that consideration.

About James Akrigg

James Akrigg is a long-time technology leader, speaker and advisor with over 30 years’ experience in the technology field, 18 of which were spent at Microsoft helping businesses of all sizes to grow and gain success adopting the latest technologies. Today, James works on the latest AI and developer tools to create digital products and uses his expertise to help businesses of all sizes to accelerate change with technology.