VENDORiQ: To Boldly Go Where No AI Has Gone Before… OpenAI’s GPT-4 to be Released This Week

What It means

OpenAI and Microsoft already have several AI models that can interpret images and video, and perform translations, in addition to the natural language generative capabilities of ChatGPT. It appears that GPT-4 will not only uplift OpenAI’s large language model but will also bring other AI models into a unified interface.

GPT-4’s multimodal capabilities will represent a significant step forward in the use of AI, bringing multiple models together in a seamless way. Think how characters in StarTrek are intact with computers, and it is possible to understand where this type of user experience is heading. That is a bold statement, but we are bold going where no one has gone before with AI at the moment. 

Microsoft has emphasised that GPT-4 is not intended to replace jobs, but rather to simplify or enhance work by adopting AI. 

Who Needs to Know

• Digital workplace teams
• Strategy teams
• CIO, CDO, and innovation teams.

Next Steps

• The release of GPT-4 is expected to have a significant impact on the use of generative AI, providing new possibilities for the interpretation of text, images, and videos. As such, companies are recommended to train their employees on adopting AI in working life as traditional job descriptions change and new professions emerge due to the enrichment of new possibilities driven by AI.

Updated

With ChatGPT-4 released, IBRS has had the opportunity to review the model’s performance. An important change in the model is the increase in the size of prompts and output, allowing for more complex use cases. Immediately, a business use case stemming from this increased capacity will be around the summation of complex documents, such as FAQs and technical or product manuals, patents and so on. Initial tests suggest that the veracity (quality and accuracy) of the summation is acceptably accurate, and the ability to then process the summations into the form of conversational responses created a powerful option for customer self-service.

IBRS has already tested the potential for the ChatGPT API to be used in complex document processing chains involving search, summation and combining multiple documents into a single natural language response, and found the effort to be within the realm of most coders with API knowledge. As a result, we expect the use of ChatGPT-3.5 turbo and ChatGPT-4 to become embedded into many self-service websites very quickly. Traditional search will not be replaced, but will merge with conversation models to deliver the ‘Star Trek’ like experience. This experience will quickly find its way into mobile devices and smart audio products (headphones, glasses, etc.) and become the new normal for digital interaction. The metaverse is more likely overshadowed by the ‘audioverse.’