A year after the release of ChatGPT — and the months of hype it has fomented — it’s clear that generative AI will become the fulcrum of future business growth. It is proving to be a crucially useful technology that will get ubiquitously embedded in applications, especially when combined with adjacent capabilities such as search, analytics, and machine learning. In the same way we no longer make a distinction between going online and offline, so too will we begin to take genAI for granted as it becomes truly pervasive. Let’s look back on where we started, where we are today, and where we can expect to go in 2024 and beyond:
Privacy And Security Concerns Were Loud, But Data is What Will Drive Differentiation
- Then: ChatGPT’s release was met with tremendous excitement but also tempered by those who attributed the way ChatGPT worked to every other large language model (LLM). In response, many organizations initially fixated on building a ChatGPT-like chatbot that would interact directly with customers, causing anxieties around data privacy and security. This between public-facing genAI products and business-ready solutions slowed the progress of many organizations.
- Now: Today, organizations realize that using company data to ground and influence the behavior of genAI applications is a crucial way to differentiate. We’ve seen the widespread adoption of approaches like retrieval augmented generation (RAG), prompt engineering, and instruction tuning to focus model behavior within the context that it’s being used. This is allowing organizations to leverage genAI in many other ways.
- Next: Most of the models available today are trained on broad datasets and can perform a wide variety of generalized behaviors. Training data and model outputs will also emerge as a practical and cultural battleground, with disputes both between and within organizations around model behavioral bias.
Use Cases Went From Supporting Individuals To Enabling A New Circulatory System For Knowledge
- Then: The first half of 2023 saw many use cases focused on how individuals can be more productive in their writing, content creation, or coding tasks especially across marketing, sales, and development teams. As genAI excitement pervaded other parts of the organization, many users took on adopting genAI on their own, raising the issue of bring-your-own-AI (BYOAI).
- Now: Today’s genAI use cases are moving beyond individual augmentation and to reach farther and deeper into the organization to connect organizational knowledge. As these applications access more knowledge, they are interconnecting individuals and teams to enable better collaboration not only between humans, but also between humans and machines .
- Next: In the future, teams comprised of both humans and machines, operating together, will change how work gets done and how they collaborate. In other cases, bots will connect to other bots. For example, calendar bots can connect with other calendars in Microsoft 365 Copilot to choose a time for individuals to meet. GenAI will play a central role as a connector, a user interface, and a pervasive member of human/machine teams.
Language And Text Took The Forefront in 2023, Image and Video Will Be On The Upswing for 2024
- Then: Image generation models such as DALL-E and Stable Diffusion opened up a new world of possibilities for visual creation, along with a plethora of challenging questions around copyright and IP, ethics, and risks of using these models. This has partially led to AI image generation taking a back seat to text generation in the past year. A steady trickle of lawsuits around the training and generation of content from both image and language models further compounded adoption.
- Now: Today, Adobe is the only image-generation model provider who can credibly claim that all the data used to train the model is owned or fully licensed by it to do so. The platform accelerates the iteration speed of the content creation workflow for those who can use it. But with this lack of choice, we are only barely beginning to see the transformative effect on content creation workflows. Even with the uncertainties around using content generated with models like Stable Diffusion, providers like Amazon have recently started offering AI-generated images for advertisements.
- Next: Many genAI model providers announced that they will indemnify their users for copyright claims against content generated by their models — but this indemnification has largely only extended to LLMs. We expect it to extend to image or video generation as well. As genAI-created or augmented images combine with text, we will see content creation speed up by orders of magnitude — a massive boon for many businesses, prompting a new wave of “white glove” premium products and services that are created or delivered by a human.
The GenAI landscape is vibrant and self-sustaining, and because of the cultural zeitgeist that ChatGPT created, continuing progress in genAI is not dependent upon the success of any single individual or organization. While 2023 was a year of excitement, experimentation, and the first stages of mass adoption, 2024 will be a year of scaling and optimization, and solving the hard problems differentiating yourself as a superhero in a world where everyone has superpowers. To dive into how to create your unique genAI advantage, read our new report “The Generative AI Advantage” and then schedule an inquiry to get into deeper questions specific to your needs.
