The tech industry and even the general press have been discussing the concept and early implementations of AI for a long time, but real-world examples have been relatively hard to find. To be clear, AI-powered functions have been in use for several years in things like computational photography on smartphones as well as high-end cloud-based data analytics applications. However, many of these are essentially invisible to most people. This year, however, both consumers and businesses should expect to see many blatant examples of AI-powered capabilities in their day-to-day activities. The launch and huge splash made by the Dall-E image generating tool (and others like it) and the ChatGPT autonomous chatbot at the end of last year started hinting at where things were going. I expect to see significantly more high-profile examples like these in 2023. To put it succinctly, while it’s been predicted many times before, I expect 2023 to be the year where AI finally goes mainstream. What’s fascinating about both Dall-E and ChatGPT is that they are examples of what’s referred to as generative AI – or the ability to create original content, information (or even software code!) from a very limited input. Also see: How to Run Stable Diffusion on Your PC to Generate AI Images With Dall-E, for example, you can ask it to generate images in various styles from a simple text description of things you want included (e.g. draw a red Porsche driving through Northern California wine country during a snowstorm in a pointillist style). In the case of ChatGPT, the capabilities are even more extensive. You can, for example, ask it to create original song lyrics or a bedtime story or even a college essay (a huge new concern for educators worried about cheating) on whatever topic and in whatever style you request. Imagine, for example, the ability to immediately generate an article describing how OLED TVs work in the stream of consciousness style of author James Joyce that incorporates characters from the Harry Potter book series. None of the generative AI tools are perfect, and many of their more egregious flaws are starting to be documented. However, they are both good enough – in fact, astonishing in many ways – that they can provide a much clearer view of what’s possible with AI to almost anyone. Plus, as with most technologies, new improved versions area already in the works—ChatGPT4 and enhanced versions of other tools are expected to appear this year.
The other interesting thing about these types of tools is that their application goes well beyond their primary functions. It turns out one of the more interesting applications for a tool like ChatGPT, for example, could be a complete re-thinking of search engines. In order to enable its incredible creative power, the software algorithm powering ChatGPT was fed most all of the text-based information available on the internet and then trained to build links across all this information. Current search engines are good at finding the information related to a request, but not good at combining it all together in a way that makes sense to normal human beings. So, instead of having to individually check out different links when doing a search on a topic, imagine being provided with a summary of the most relevant information. That type of AI-powered search takes it to an entirely other level. Recognizing this, Microsoft is rumored to be bringing some of the ChatGPT functionality to an updated version of the Bing search engine – as early as March. (Microsoft is a significant financial contributor to OpenAI, the company that developed ChatGPT and Dall-E.) In addition, red flags have apparently been raised at Google about the potential impact of these types of tools, so many are expecting a combination of the company’s own LaMDA, MUM and PaLM generative language models to be incorporated into a future version of Google Search. The significance of AI applications becoming commonplace and impactful for a large percentage of the population goes well beyond the abilities themselves, however. As with many technologies that essentially hide in plain sight before they’re widely understood and appreciated, these mainstream AI capabilities will open people’s eyes to the many other types of AI-powered functions that currently exist (and continue to be improved). Everything from the ability to automatically generate photorealistic background images from simple sketch drawings, as Nvidia is doing with their Canvas tool, to the ability to improve 5G signal reception in smartphones through modems that “learn” how to adapt to the signals in a given region, as Qualcomm is doing with their X70 chips, AI-powered “background” tasks are going to start getting more recognition and appreciation this year. As with many technologies that essentially hide in plain sight before they’re widely understood and appreciated, these mainstream AI capabilities will open people’s eyes to the many other types of AI-powered functions that currently exist (and continue to be improved). In addition, the need to have devices that can help accelerate AI-powered software functions will be recognized as becoming essential. At CES 2023, for example, Intel and AMD demonstrated new hardware acceleration capabilities for AI-based software in their latest CPUs. Intel is incorporating technology from their Movidius line of AI accelerators into certain versions of their 13th-gen Intel Core processors and AMD unveiled a whole new AI architecture for some of their newly announced Ryzen 7000 chips. Dubbed Ryzen AI and featured in the new Ryzen 7040 line, the technology incorporates a dedicated engine designed to speed up AI applications. Both Intel and AMD efforts are akin to the early days of integrating GPU graphics engine into their CPUs. Expect many more developments on this front in the PC world because of this. The bottom line is the spillover impact of mainstream AI tools like Dall-E and ChatGPT is likely to be as important and probably even more far-reaching than they are. As a result, expect to see a great deal more focus on AI-related capabilities in products, software and services of all types throughout the new year. Bob O’Donnell is the founder and chief analyst of TECHnalysis Research, LLC a technology consulting firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on Twitter @bobodtech.
