The Netscape Moment.

OpenAI’s release of ChatGPT was to AI what Netscape’s browser was to the Internet. They both democratized a powerful existing technology. What was once the purview of academics and computer scientists suddenly became a mainstream tool. Existing technology companies scrambled to incorporate the Internet into their products just like technology companies are scrambling to incorporate Generative AI today. In the early days of the Internet there were sceptics… there always are sceptics. In 1995, Robert Metcalf, inventor of Ethernet said, “I predict that the internet will go spectacularly supernova, and in 1996 it will catastrophically implode”. We bet he wishes he could take that one back. Generative AI has its sceptics too. At Infinitive, we’re very optimistic about GenAI. We rather like the take Elon Musk has on the topic, “Generative AI is the most powerful tool for creativity that has ever been created. It has the potential to unleash a new era of human innovation.”

Let’s talk about how Generative AI will evolve for businesses over the next three years.

2024 — Turbocharged chatbots, virtual assistants and the return of shadow IT.

1. Chatbots on steroids. Generative AI using Large Language Models (LLMs) will be embedded in more than half of today’s conversational applications (in 2023, it is in 20% of such applications). This will dramatically improve the current generation of chatbots. Customer service will be the biggest beneficiary inside enterprises. From phone trees that customers don’t hate to SaaS application tutors that start to take the place of Customer Success Reps, the world of talking to technology will get much better. Many of these GenAI-enabled chatbots will be incorporated into existing services. Airbnb’s website might allow a customer to initiate a conversation where rental properties are considered, restaurant reservations are made, tickets to sporting events are purchased, etc. NASCAR could allow somebody attending one of its races to find convenient motor home parking and local fishing guides. All with conversational natural language. Companies who seize the initiative early are likely to maintain their lead
2. Digital Virtual assistants come of age. Technology companies have been trying to perfect the digital assistant since at least 1993 when Apple introduced the Newton. 2024 is the year that virtual assistants cross the chasm. Generative AI reboots of existing virtual assistants, like Assistant by Google and Alexa by Amazon, are already in the works. The big step forward in 2024 is that these Generative AI enabled assistants won’t just retrieve information, they will take action. A typical conversation might include asking the assistant about the NFL’s schedule, finding a ticket on the sunny side of the field for a game, buying the ticket and scheduling an Uber to take you to the game.
3. The shadowy world of shadow IT. The big Generative AI vendors will continue to improve their base products. As those products allow users to upload documents, diagrams and the like, employees will use their personal accounts for business. Examples include an internal memo edited by ChatGPT for clarity or a webpage rewritten for SEO optimization. Corporate IT leaders will either provide their employees with a sanctioned Generative AI capability or take the chance that proprietary information will leak into the public domain through the unsanctioned use of personal accounts.

2025 — Co-pilots and the rise of AI TRiSM (Trust, Risk and Security Management)

1. AI augmented software engineering will be the primary focus of co-pilots in business. A generative AI co-pilot is an artificial intelligence system that assists users by generating content, suggestions, or solutions based on the input it receives. The term is frequently used in Generative AI to describe a tool to help developers write software code. Products like GitHub Co-pilot are already providing productivity improvements for developers.  As the models get better trained and the development co-pilots become more integrated into DevOps toolkits, almost every professional developer will be using a co-pilot. Microsoft is set to release its Office co-pilot on Nov 1. It will be will be embedded in the Microsoft 365 apps including Word, Excel, PowerPoint, Outlook and Teams. By 2025 just about every enterprise business application will have some level of co-pilot functionality.
2. In tools we trust. The policy and security issues accompanying the rapid deployment of Generative AI in 2024 will come home to roost in 2025. This will generate interest in a new area of IT management named Trust, Risk and Security Management (TRiSM). Tools for activities like model interpretability and explainability, content anomaly detection, AI data protection and more will be required by 2025. Efforts are in place to develop such tools today by startups like AIShield, Fiddler, ModelOp, and SolasAI.

2026 — Retrieval Augmented Generation starts, self-supervised learning appears and model hubs predominate.

1. Retrieval Automated Generation (RAG). RAG is a relatively new artificial intelligence technique that can improve the quality of generative AI by allowing large language models (LLMs) to tap additional data resources without retraining. In a corporate setting, data is constantly changing. Incorporating the new data into the LLM would require constant retraining of the LLM. RAG allows for a hybrid “search” and Generative AI process with the new data merged with data from the LLM. This will allow enterprises to securely combine internal corporate data with LLMs.
2. Self-supervised learning (SSL). Self-supervised learning is a type of machine learning where systems are trained to predict or solve tasks using only a raw dataset, without the need for human-annotated labels. The ability to learn directly from raw data significantly reduces the need for extensive, costly, and time-consuming data labeling processes.  The advent of self-supervised learning will open many more datasets to AI.
3. Model hubs. Model hubs are repositories that host pretrained and readily available machine learning (ML) models, including generative models. They also offer automation and governance tools, curated datasets, model APIs and generative AI-enabled applications targeting specific enterprise needs. These “AI supermarkets” will facilitate the creation and sale of domain-specific AI models that apply to specific industries or sub-industries. These domain-specific models will allow enterprises to get even more value from Generative AI.

How Infinitive can help. Every enterprise needs to start thinking about how it will apply Generative AI to its business. The revenue uplift of incorporating Generative AI into products and the cost reduction of automating previously manual tasks are too big to ignore.

What is your company’s AI adoption plan?

Infinitive works every day with customers using Generative AI models and applications.

Contact us to talk about how AI can improve your business.