top of page
Writer's pictureWilliam Webster

The Low Hanging Fruit of AI

Updated: Jan 11

It seems that everyone is utilising artificial intelligence. However, the narrative often contrasts sharply when I engage in personal conversations. For instance, I recently learned about a bank investing millions in developing models. Yet, when talking with its employees, many claim they aren't using or even aware of such tools. It's astonishing to see such a sophisticated strategy that overlooks the basics. Large Language Models (LLMs) like GPT-4 are democratising AI, granting anyone the potential to enhance their tasks — the only requirement is the willingness to embrace it.


My work habits have changed. I used to search online or ask colleagues for answers. Now, I often turn to Large Language Models (LLMs) like GPT-4. This change comes from trying them out and seeing their efficiency. They save time and often give better results. So, I wonder, why aren't more businesses using tools like this?


  • Inertia: Human nature tends to favour the familiar. We often gravitate towards tried-and-tested methods, which can prevent us from fully understanding or appreciating the capabilities of new innovations.

  • Time Constraints: Our existing commitments often demand immediate attention, leaving little energy or inclination to venture into the unfamiliar territory of new tools and technologies at the end of the day.

  • Turf Wars: Technology adoption can often be perceived as an IT-centric endeavour. This perception can lead to extensive processes like planning, development, testing, roll-out, and sign-offs. Solutions that bypass these stages can introduce inter-departmental friction.

  • Access Restrictions: Most workplaces have strict protocols around software installations and gaining permission often involves multiple layers of approval, making spontaneous adoption challenging.

  • Security Concerns: Data is a valuable asset. The threat of breaches or leaks can deter companies from trying new tools. Unauthorized access can benefit competitors and prove costly and damaging to a company's reputation, especially for regulated entities.

  • Misinformation Risks: LLMs, though powerful, can sometimes generate inaccurate or misleading narratives. Associating your name or brand with such content can have unintended consequences.

Even with these challenges, it's important to note that progress has its ups and downs. With the right planning and approach, we can overcome these obstacles and pave the way for new ideas and growth.


Many businesses already rely on third-party services for data management, so outsourcing is not a novel concept. There's an underlying principle of mutual interest at play. Service providers are just as invested in preserving the integrity of the data they manage. Any breach, misuse, or unauthorised copying of client data would jeopardise their own business model and reputation.


Do LLMs occasionally generate incorrect, biased, or misleading outputs? Certainly. But this doesn't render them unreliable. We have safeguards in place, one of which is human oversight. Just as we review spreadsheet results for accuracy, we can and should scrutinise LLM outputs to ensure they align with our expectations. Blindly accepting any tool's output without questioning or validation can lead to issues down the road.


It's disheartening to see many businesses overlooking the potential of AI. In doing so, they also disadvantage their employees. As AI becomes increasingly integrated into industries, professionals unfamiliar with its capabilities risk being outpaced by those who can use it.

Deploying these models can yield immediate benefits, and the investment of time and effort is minimal. A structured approach, spearheaded by a board-empowered project sponsor, can streamline the integration process.


Here's a simple plan:


  • Sign Up: Register for GPT-4 or its enterprise version.

  • Empower the Workforce: Distribute access among your employees and encourage them to experiment.

  • Feedback Mechanism: Set up a centralised system where users can document their experiences, uses, and feedback. This will provide invaluable insights into the model's real-world applications.

  • Training: Although intuitive, a brief demonstration on how to effectively prompt the model can save hours in the long run.

  • Review & Adapt: After a few weeks, review the innovative applications your team has discovered. The financial incentives alone, such as the time saved in drafting documents, make this endeavour worthwhile.


The proposition is straightforward and cost-effective. Why not take the leap?


Getting Started with GPT-4


While GPT-4 doesn't come with a manual, here are some steps:


  • Registration: Opt for the paid version, priced at $20/month for individual access

  • Switch it on: Navigate to settings and grant yourself access

  • Experiment: Use the prompt window to test various queries.

  • Toggle Features: Explore different modes to familiarise yourself with the model's capabilities.


From here, it's all about learning through experience.

23 views0 comments

Recent Posts

See All

Comentários


bottom of page