5 predictions for Generative AI in 2024

Ed Vickers

June 13, 2024

Generative AI

In the past year, generative AI has transitioned from a fascinating novelty to an integral player in the tech arena. Now in 2024, we’re eager to unearth the practical applications set to redefine industries, especially within financial services and other regulated environments.

In this blog we forecast the imminent evolutions based upon our intimate work with clients and experts in the field. Here's a peek into the future, focusing on how generative AI's growth may unfold in the coming year.

1 - A move from generalist to specialist

While there’s been a trend toward larger language models (LLMs), we’re seeing a shift toward developing smaller, more specialised generative models tailored for specific tasks, particularly in highly regulated environments like financial services. Loop Agencies are already part of consulting groups with some of our clients on bespoke LLMs that are trained to write marketing copy/content in a specific tone of voice, with all the relevant FCA and FSA guidance and rules in place.

It’s still a long way from being commercially viable and, at the moment, with a lot of human intervention along the way, but as this evolves, it will be a solution for large-volume communication. This could help address concerns about computational resources and ethical considerations and we can expect to see a focus on refining and developing more effective fine-tuning techniques, allowing developers to adapt pre-trained models to specific applications with greater ease and efficiency.

2 - Emerging Experts - Development at pace

We’re seeing more and more experts emerge. This will only help our understanding and adoption of AI for businesses. The real key will be finding experts in particular areas, finance, for example, who understand markets, audiences and products and who also understand AI and its application to that specific need. A person with knowledge and expertise on a subject will always get far better results from AI than someone with no previous expertise.

3 - More governance

As generative AI becomes more prevalent, there may be a greater push for the development and adoption of ethical frameworks and guidelines to ensure responsible and transparent use of these technologies.

There also needs to be an increased focus on making generative models more interpretable and explainable, addressing concerns about the "black-box" nature of these systems and facilitating trust among users.

Researchers and developers may work on enhancing the robustness of generative models to adversarial attacks and developing security measures to prevent misuse, such as generating malicious content or deepfakes. This still remains one of the biggest concerns with generative AI and will remain so for the foreseeable future.

4 - Commercial models/licensing

As Generative AI expands and develops, we see a point at which corporations who hold vast amounts of research and data will license this data to large language models, providing a service where information is managed and distributed, potentially in different tiers based on a subscription model. Making some Generative AI tools smarter than others. For LOOP Agencies, who work in the Financial Services and Regulated Services space, governing bodies and regulators may charge to have their information included in corporate LLMs

When it comes to copyright issues, we’ve seen a few high-level cases widely reported in the news. Governments and laws will be expected to do more to provide clarity around what’s deemed as copyright infringement going forward, particularly in the imaging space. Already, we’re seeing paid-for services by all the large stock libraries including AI content.

5 - Prompt engineering will become even more important

Previously, we’ve discussed prompt engineering, and as these AI models become more refined, a prompt engineer’s knowledge of that specific AI model and how to optimise the results will be key.

Strong prompt engineering is crucial for unlocking the full potential of generative AI models and achieving accurate, high-quality results. It's like the conductor in an orchestra, guiding the model's vast knowledge and abilities towards the desired outcome. Here's why I think strong prompt engineering matters:

Precision and Specificity:

Imagine throwing a wide net when fishing. You might catch something, but it could be anything. Vague prompts are like that. They give the model too much freedom, potentially leading to irrelevant or inaccurate outputs. A strong prompt, however, acts like a targeted lure, providing clear instructions and context to focus the model's attention on the specific task at hand. This enhances the chances of accurate and relevant results.

Controlling Creativity:

Generative AI excels at creative text formats, like poems or scripts. However, uncontrolled creativity can be messy, particularly when it comes to image models. A skilled prompt engineer can nudge the model's creative output in a specific direction, shaping it to fit the desired tone, style, or theme. This is vital for tasks like writing marketing copy, creating imagery or generating code in a particular language.

Mitigating Bias and Misinformation:

Unfortunately, biases and misinformation can exist within AI models. Prompt engineering allows us to address these issues by providing counter-examples and factual information within the prompt itself. This helps steer the model towards more equitable and accurate outputs.

Strong prompt engineering is the bridge between the raw power of Generative AI and its real-world application. A Prompter with knowledge and experience of a particular subject will still far outweigh a Prompter with little to no knowledge.

It's not just about getting a result. It's about getting the right result.

Effective prompting creates an ongoing dialogue between the user and the model. The more you can experiment and refine your prompts, the closer you'll get to achieving the desired outcomes.

And just for the record, no, we didn’t ask ChatGPT to write this for us... But hold on, there's more to consider when it comes to the generative AI ecosystem. It's clear that the careful orchestration of innovative technologies will continue to be central to strategy, especially within the confines of financial services. With in-housing and outsourcing of creative campaigns rethinking traditional workflows, and financial services seeking fresher avenues to engage audiences, the role of agencies is being redrawn.