Predictions re: Artificial Intelligence and Machine Learning

The internet is positively ate up with all of the buzz surrounding artificial intelligence and machine learning. Generative AI is dazzling us with creative potential. Machine learning is wowing us with its promise to help us slug through gobs of data and make sense of things. Big companies like Microsoft, Amazon, Google et al. are betting big on AI to transform entire industries optimize their cash flow, and yet no one really knows how things are going to pan out.

In this post, I intend to add my own personal speculation to the mix.

To date, I’ve tried a few AI/ML tools including:

  • ChatGPT (OpenAI), which helped me by writing code for a large scale information architecture project I completed a few months ago. It was my first experience with the concept of prompt engineering.
  • Bard (Google), which seems to work pretty well, although I’m not sure if it’s quite as polished as ChatGPT
  • DALL-E (Open AI), which generated a bunch of images that felt like hallucinations—nothing nearly as slick as the marketing images.
  • Firefly (Adobe), which I’ve attempted to use for creating blog cover images. I haven’t used previously generated images here because they didn’t match the intended concept and/or fit the aesthetic, but the implementation of Firefly in Adobe Express is really cool.1The cover image for this post was generated using Adobe Firefly in Adobe Express. Prompt: “machine learning robots in home” I selected the “Pixel Art” theme. Plus, it provides several options based on text input and it allows you to pick an artistic style for the output. Those controls make it a great tool for generating ideas and concepts, even if the output is really bizarre.
  • Whisper (OpenAI), which I used to transcribe a 14 minute speech. I wanted to use it because it’s a locally-deployed command line tool. It did a fantastic job of accurate transcription, although I wish the documentation had been a bit better so I could understand the options for commands a bit better.

And beyond the ones that I’ve directly interacted with in the manner described above, I know that I’ve encountered AI/ML tools in various contexts for years. The fact that such tools exist is not new. What’s new is the power, speed, scale, and sophistication of emerging AI/ML tools. They are here to stay, and before the dust settles, I’d like to make a few predictions about where things are heading.

My Predictions on AI/ML

Full disclaimer, I’m not an AI/ML expert. I’m just a regular guy who works with computers and the World Wide Web. My predictions are rooted entirely in my perspective and also some historical analysis.

  1. Privacy, data security, and misinformation concerns will generate increasing demand for local/on-premises AI/ML tools. Unfortunately, no one really reads privacy policies because—let’s face it—most of us aren’t lawyers. But I think there is a growing unease about whether the data collected by using an AI/ML tool will be used to train the tool. Issues related to fair use and intellectual property get pretty weird real quick when it comes to how LLMs are trained. I can see a lot of cases where an organization might see tremendous value in using an AL/ML tool, but the concerns about data privacy will push them to find solutions they can run on their private machines or servers.
  2. More and more on-premises/local AI/ML tools will become available. Three current examples include Synology Photos, which uses local ML to categorize photos on a your private server, Synology’s Deep Learning NVRs, which features on-premises AI video analysis, and on-device speech recognition on iOS and MacOS devices. I can envision a future where AI/ML tools can be installed as easily as a WordPress plugin or a package via Homebrew and run locally.
  3. The open source community will develop plug-and-play AI/ML tools that can be modified for specific use cases. This prediction simply extends a current process that is already taking place. Tools like Whisper and TensorFlow are already open source. My thought is that we will see a rise of smaller open source projects that solve specific problems. Independent devs and corporate teams build specific software solutions for their problems and put them out into the world for public use and modification. Why would AI/ML tools be any different?
  4. As people gain a greater awareness of AI/ML tools, we will see more niche applications developed and deployed. The vast array of WordPress plugins, web frameworks, and command line tools that are available underscore the fact that there are many problems to be solved and many different ways to solve them. Coding is practical problem solving using the tools of the trade. AI/ML are powerful tools in their own right, and I think different industries will use them in unique ways. Whereas ChatGPT is a general purpose LLM, GitHub CoPilot is a niche application. I think we’ll see more of those tailored applications of ML/AI for specific business and organizational problems.

I might be right, or I could be completely off the mark. Maybe we end up in a real life version of Terminator or Wall-E. Most predictions about the future tend to be off base because they account for just a fraction of what’s happening in the world. I’m not sure what other developments could aid or disrupt the future of AI/ML tech: Will there be breakthroughs in chip design? Will governments crack down on AI/ML in an unexpected way? Will Wreck-It-Ralph break the internet?

Only time will tell. I don’t think these tools are going away any time soon, so I think the central question is: What will they look like in the future? I’m aware that my predictions may seem a bit vague or non-specific. I’m not here to make audacious claims. I’m simply tracing current trends and trying to anticipate where they might lead. If the trend towards private clouds and data ownership continues, I think there will be a unique application of AI/ML there. And I think the open source community will simply do what it does and crank out neat tools and proudly post them on GitHub for the world to see and benefit.

When the future happens, we’ll see it happen at the same time.

  • 1
    The cover image for this post was generated using Adobe Firefly in Adobe Express. Prompt: “machine learning robots in home” I selected the “Pixel Art” theme.
Scroll to Top