GPT in practice

- While many people will use the ChatGPT service directly on the web or through its own app, its backend large language model, or LMM, is a software platform that enables all types of innovation to be built on top of it. By this I mean that GPT technology can be incorporated seamlessly into existing solutions via application programming interfaces, or APIs, software that enables two or more computer programs to communicate with each other. GPT could also be the basis for completely new innovation. For example, Microsoft, a major investor in OpenAI, is quickly building GPT capabilities into many of its products, including its popular Office suite and its search tool Bing. Under the moniker Microsoft 365 Copilot, these GPT-based functions will enhance and simplify existing capability and also add completely new features. For example, in Outlook, a user will be able to get a summary of emails in a long thread rather than having to read through all of them. As an example of new innovation, a service called Be My Eyes that has typically used human volunteers to describe images to visually impaired people is beginning to use GPT to automatically process visuals and then describe them without any human intervention. Further still, this new GPT-powered version enables a user to ask AI questions about what is in the image. A variation of this supports the ability for a person to take a picture of the contents of a fridge and then ask GPT to suggest recipes that would use the ingredients it can see in the image. This capability can also be used to translate written language by simply taking a photo of, say, a food product from a foreign country and having GPT process the text and produce output in the language of choice. The popular online and free educational service from Khan Academy has being quick to adopt GPT capabilities. They use AI as a teaching aid to help students arrive at the correct answer to a question or to better grasp a complex topic. For example, an area they're developing is the ability to interview historical figures such as Cleopatra. Imagine that. For teachers, AI is helping to create lesson plans and assisting with administrative tasks. For developing software, GPT is having a profound impact. Existing developers can paste problematic code into ChatGPT and it will find the bug and suggest the remedy. Code can also be generated in the programming language of choice simply with a plain language prompt. Creating a complex SQL query becomes a breeze. Even more remarkably, a picture drawn of a desired app screen or a simple text description can be processed into the code for a basic functioning app. Additional features including elaborate graphical user interface design is simply a few prompts away. Once the domain of experienced developers and designers, because of GPT, writing code or no code development is becoming far more accessible. GPT can help to process a wide variety of business-related requests, including the generation of ideas, image creation, reports, math calculations, presentations, stories and scripts, curated lists, emails, and content summaries and explainers. It can also simplify complex text, do translations, compare and contrast items, and conduct research. Requests can be combined where, for example, GPT can be asked in one prompt to search and find information, simplify it, and present it in text or visual form in the style of your choice. One solution called AutoGPT when only given a goal instead of explicit requests will create the relevant tasks to achieve that goal and then execute them. Watching the software determine the right tasks, sometimes changing its mind based on results, and then creating a new task is magical to watch. Finally, an area that is being transformed by GPT is customer service, historically a human-intensive exercise. Attempts to automate the response to questions, requests, and complaints has only been marginally successful in the past few decades. More often than not, customers get frustrated and have to speak to a human. Now with GPT-powered AI, customer service may finally be ready for prime time. Conversations are human-like, responses are accurate and intelligent, and video and voice can be incorporated. It won't be long before customer service with AI is commonplace, and it may even be preferable in many instances to working with a human. A challenge in providing examples of GPT in practice is the speed at which new AI-driven innovation is emerging. While my examples in this video give you a flavor of the many ways GPT is being used, by the time you watch this, there will be many more incredible solutions available. After you watch this video, ask ChatGPT to tell you all about them.