Hong Kong EN

Balancing AI Expectations: Navigating the Journey from ‘Good Enough’ AI to ‘Great’ ROI AI

Ryan Cox

Senior Director and Head of Artificial Intelligence , London, UK

Artificial Intelligence

The perfect is the enemy of the good.

Other than true perfectionists, we humans often implicitly follow this maxim. Even those who don't apply it to themselves will usually tolerate ‘good enough’ in other people.

When it comes to computers we expect mathematical perfection – because that's the idea we’ve been sold.

But, is the pursuit of perfection hindering the practical benefits AI can deliver today?

The computer says no... is an unarguable statement

Even in making mistakes, we attribute perfection to these machines: to err is human, to really mess things up requires a computer.

We all know that computers aren't perfect, and yet we rely on them all the time for our communications, our banking and for the management of our daily lives. We've accepted many of the flaws of technology systems that don’t use AI; however, we’re still intolerant of flaws in AI itself – because the hype has given us the impression that adding a dash of AI makes the system (finally) perfect…

But, at Synechron, we have a different view…

We believe that success requires human + AI. The journey from ‘good enough’ AI to yielding ‘great’ innovation and productivity lies in the synergy of human creativity and algorithmic precision.

What machines can do for us is reduce the mundane – the repetitive, boring work that makes up the 99% perspiration required to deliver the 1% inspiration. If we can reduce this toil, it means we humans can do far less – meaning our jobs become more enjoyable and our productivity increases.

AI can help us in a number of key ways…

Business objectives

First up, naively, I drafted my work objectives to align my goals for the next quarter. It took about 30 minutes and the tone was good. Then I wanted to put AI to the test: I turned on my speech-to-text recorder and spoke for about 2-3 minutes on what I wanted to focus on and achieve; then I asked Gemini to produce the equivalent set of objectives… it took 30 seconds. After reading it, I had to make a few changes to get the sense right, add a few additional specifics, and correct some awkward grammar – but then I was ready to double down on my goals and move forward – in a grand total time of around five minutes. I wouldn’t have been able to use the draft AI objectives as it was, but it was good enough that I could (after a tiny bit of human effort). So, Human + AI made a big difference.

Pitch decks

Our new Synechron Nexus AI product, Synechron Amplifai, can produce an easy-to-navigate pitch deck. In minutes, it creates a set of PowerPoint slides from many web pages of publicly available information across the internet. The data sources are configurable, so companies can plug in their own data sets as needed. Gathering, summarizing, and visualizing the information in charts might previously have been undertaken by a junior banker – potentially saving several weeks of manual effort. The output isn’t ready to be pitched immediately, but it contains references back to the source material and is ‘good enough’ and ready for a human to review, and finalize. This frees them up to use their brain power on strategy and creativity, as the mundane work of pulling data has been automated.

Policies

GenAI can be a critical tool in maintaining organizational integrity, whether it’s for internal policies, such as for HR, or for external ones, such as those from a regulator. Organizations deal with so many of them that having a system which can parse through and identify highlights, or respond to a specific question is very useful.

At Synechron, we’ve found that GenAI can take on most of this job, providing a generally ‘good enough’ answer to most policy questions, along with a link to the original policy documented. When we did a deep dive into why AI was failing to provide a proper summary, we found that it was the humans that had failed (when a few policy documents needed to be worded in a clearer manner – for human and AI consumption). As policies are constantly being updated and added to, this process will continue, with both the human and the AI working together to produce answers.

Translation

Even before the advent of GenAI, ‘good enough’ tools have existed, such as Google Translate. These can translate mainstream, non-colloquial sentences with little problem. GenAI has now taken this to a new level, providing better accuracy and a wider set of language combinations. However, it’s usually quite easy to spot when a machine translator has been used, which is why translation companies often use these tools to do the first pass (the mundane), and then use humans to read and polish, and make it read naturally with the right tone and context for the situation. For day-to-day translation, AI is already good enough but, for perfection, we still need human beings.

It's about making the most of AI now

As AI continues to evolve, its capacity to absorb and streamline mundane tasks will increasingly transform industries, elevating 'good enough' AI to 'exceptional' AI. However, the true measure of success lies in how AI complements and amplifies human ingenuity, enhancing our creativity and decision-making processes.

At Synechron, our mission extends beyond merely harnessing incremental advancements in AI. We aim to orchestrate a human-AI partnership to unlock productivity and innovation, helping businesses achieve extraordinary ROI from their AI investments.

As we navigate the journey from ‘good enough’ AI to ‘great’ ROI AI, we’ll continue to push the boundaries and ensure that AI serves as a catalyst for human potential, reducing the mundane, driving innovation and propelling our clients’ success stories in the digital age.

The Author

Rachel Anderson, Digital Lead at Synechron UK
Ryan Cox

Senior Director and Head of Artificial Intelligence

Ryan Cox is a Senior Director and Synechron’s Co-Head of Artificial Intelligence. We partner with companies to explore the potential of AI technology to revolutionize their business. Synechron's AI practice specialises in large language models, generative AI technologies, AI strategy and architecture, and AI research and development. We ensure AI systems and solutions deployed at our clients' sites are ethical, safe and secure. Contact Ryan on LinkedIn or via email

See More Relevant Articles