Platforms like ChatGPT can be put to work on a mountain of repetitive coding tasks, Writes SIJU MAMMEN technology lead at Elenjical Solutions.
In a recent pilot study, Elenjical Solutions, a leading South African fintech company, tested the capabilities of various Language Learning Models (LLMs) in multiple scenarios and measured an increase in productivity while using these tools. In fact, in certain cases, software developers using these tools were able to write complex computer programs in about half the time.
I have been closely following AI’s development since 2017. There was a feeling within the tech community that it would take decades for AI to reach human level natural language communication. Incredibly, however, it has only taken five years.”
Force behind the future
Much of the excitement surrounding AI can be attributed to its ability to facilitate natural language communication with technology. As humans, the most intuitive way we communicate is by talking to one another. And so, the prospect of machines being able to ‘talk’ with us intelligibly and display cognitive behaviors is the holy grail of human/machine interaction.
AI is not one piece of tech, but rather a collection of tools, techniques, algorithms, and data working together to achieve a certain goal. ChatGPT, for example, is built on all the knowledge of the world we can find on the Internet. All of this information has, in a way, been internalised by these tools.
Machine learning is one of the most common types of AI in development for business purposes today. It excels in processing vast amounts of data rapidly. These AI algorithms appear to ‘learn’ and improve over time. Integrating AI into any organisation is a substantial undertaking, especially within niche businesses. It takes in-depth knowledge and a lot of time and effort to train the system and iron out inaccuracies.
Elenjical Solutions is famous for its deep knowledge of the Murex financial technology solution and tools such as Chat GPT will over time be able to handle many tasks efficiently and accurately. However, it is essential the AI is securely underpinned by deep human expertise.
AI tools have become surprisingly good at generating boilerplate code—not necessarily complex code, but the repetitive code that every programmer needs to write, such as frameworks for various tasks. These tools can churn out such code efficiently.
Programmers at Elenjical Solutions envision a global goal for AI development—a reality where AI can perform virtually all human tasks without distinction. However, unlike much of the industry, I believe that we are still a few decades away from achieving this. And even if this is achieved sooner, hard problems are hard, not because people are unintelligent, but because they are inherently challenging. They will still be hard for AI – but the combination of humans and AI empowered machines offers great potential for generating positive outcomes.
Working with companies that deal with significant financial transactions naturally raises concerns about data security and privacy when integrating AI. The reality is that the system does not have unrestricted access to such information itself.
It’s crucial to be careful regarding the use of private, proprietary information when using these systems. Fortunately, privacy and security concerns can be alleviated by making certain architectural decisions when creating systems that use these tools.
Accountancy firm PwC predicts that AI will increase global GDP by up to 15% by 2030, with the same research suggesting that AI could contribute as much as $15.7-trillion to the global economy by that time.
A bit like cheating schoolboys, the rise of AI in business is unstoppable. AI holds immense promise for business growth, efficiency, and innovation. However, it’s essential to remember that AI and humans are not competitors in this brave new world, but important collaborators.