Generative AI: Current State and Future State (e.g., tomorrow)

Reading Time: 3 minutes

Writing a blog post detailing the current state of Generative AI (GAI) feels like a futile exercise. Why? During the writing of this post, the state of the industry will change. So, read with caution because like GAI, I can make mistakes. 

Having dispensed with the disclaimer, let’s get into what you are looking for in this post—the current state of GAI. First, OpenAI’s ChatGPT continues to lead the GAI industry, but Microsoft and Google are making advances with Copilot and Gemini, respectively. Amazon, Mistral, Facebook, and other open source models continue to push OpenAI to innovate and produce more value, but they cannot compete with ChatGPT. Apple, one of the richest and most resourced companies on the planet, still has no answer to ChatGPT, Copilot, or Gemini.  

But Apple seems poised to make a change with a possible debut of a smarter Siri. The new M4 processor recently released on the new iPad Pro expands Apples ability to do GAI calculations. Apple skipped the release of the M3, currently powering the most recent Macbook Air, and moved directly to the M4. You may be thinking, “who cares or why does it matter?” I’m glad you asked. Neural processing cores make GAI possible. And Apple has been slowly ramping up this kind of processing in their chips since 2017 when they made Siri possible by debuting the A11 Bionic with two neural cores. The M4 performs 6,233% better with a 16 core neural engine. So, Apple may announce something at their upcoming WWDC, but for now, they are not in the GAI race.  

Second, from where I sit, I think a general distrust of AI remains. In a recent experiment here at Cedarville University, 300 students were given access to an official ChatGPT 4 Turbo bot programmed to help them study for their final exam, but only half used the tool (more to come in another blog post). While I do not havedata to support my assertion beyond this lack of use, it is at least a viable explanation. Additionally, Gallup surfaced this same sentiment in an article supported by their poll data by Jeremie Brecheisin. Although this distrust exists, there are those who are using the tool and finding it very helpful in their day-to-day. 

Third, the models are getting more powerful and cheaper. Currently, OpenAI is now on ChatGPT-4o as a refined model of ChatGPT-4 that is cheaper, faster, and just as accurate as ChatGPT-4 Turbo. They continue to add additional features such as Whisper that allows you to have a voice conversation with the model in any language. You can also have ChatGPT-4o create and analyze graphics and art work through its seamless use of Dalle-3 (OpenAI’s image generating AI model). Video is also a part of the ChatGPT 4o release, but it is currently only available to “strategic” partners for testing.  

Finally, we are beginning to see the advent of small language models (SLM) that will bring GAI to specific domain areas like Kahn Academy’s free release of Khanmigo for teachers. Khanmigo will tutor students in math and other subjects and confine itself to those domains. As a result, it will cost considerably less and be easier to refine and maintain for specific tasks. Khanmigo will not do everything, but it will do certain things really well. If you are a teacher, check it out since it is free. 

Now, some may wonder what’s next on the horizon for GAI? Well, I’m glad you asked. Along with the growth of SLMs and LLMs, large action models (LAM) will become more mainstream within the next year. LAMs are the connection of current LLMs like ChatGPT with other goods and services. For instance, imagine a Siri, Alexa, or Google that could actually carry on a conversation that resulted in the purchase of everything needed for a vacation trip. Or imagine that you are staring at a refrigerator void of whole milk because someone selfishly drank it all and left you none. Never fear, you will soon be able to ask ChatGPT or some other assistant to order it for you from Kroger to have it delivered to your door within 30 minutes. Imagine what it could do for you as a teacher or a student. 

So, what do you imagine it doing for you? What would you allow it to do for you? Drop me a line and let me know. 

Want to never miss a post? Subscribe here!