I am writing a series of blog posts that will take several predictions that have been made about AI for 2024 and work to understand whether I think they will come to fruition, what they would look like if they did, and in general try to use the science fiction writer part of my mind to see into the near future.
The emergence of Niche AI’s will dominate the Generative AI landscape.
There is no denying the open-source community and smaller companies who are taking the bones of the LLM models and creating their own niche products that cater to a specific need. I’m not talking about the writing, training, and reasoning ones that are already out there. A lot of that functionality is already in the larger LLMs and working just fine. I’m also not going to worry about the AI “add-ons” that seemingly every company is putting into their existing services. Although some of those do fit within the niche category (the one’s trained on product specific data), most currently are “bolt on” AI LLM’s with mainly chat-bot features. I’m talking about the models that people have been training to only cater to niche specializations.
These are LLMs or apps that are focused on a specific market/purpose and aren’t trying to be anything beyond that. I’m talking cybersecurity, smart home management, marketing, healthcare, logistics, education, workflow automation for specific products and markets, legal contract solutions, and on and on.
The 2024 AI landscape will be full of news about these niche AI’s and how they will transform how we work. And they absolutely will. I expect the “bolt on” AI’s in existing apps will become so basic by the end of 2024 that several companies who don’t utilize the open-source community’s work will begin to suffer as more agile startups take their business. We are looking at a disruption in several markets and jobs.
There are opportunities for people with technical backgrounds to innovate in a way that the larger LLM companies won’t even be considering. And we will all be better for it. Even those larger LLM’s will benefit from the ingenuity that individuals and small to medium companies will be creating this year and far into the next few years. Don’t think for an instant that OpenAI/Microsoft, Meta, or Google leadership aren’t closely watching these microcosm AI models that work effectively at one thing and trying to see how they could take advantage of that in their larger services/models. They have teams who are working on their own niche AI models to stretch their own boundaries. And if they aren’t, they will be left behind.
This year will be dominated with new niche AI models forcing us all to address them and the opportunities they offer or ignore them and take a risk that they are a flash in the pan. And it’s that last bit that will be interesting to me. Most will be temporary. A true niche that only people in that area will appreciate and use and some even then will see more benefit out of the larger LLMs for that purpose and others. I expect you’ll see some innovative models that will help in the short to mid-term and then be replaced by other models that achieve better results or, again, by the larger LLM’s that can now do those innovative things and so much more for you.
The same reason that you’re seeing the multitude of Niche AI models and services appear (the relative ease of creation) will also be the reason that we will see them replaced or grow into something larger than they can sustain.
I am not saying that 2024 won’t be the year of open-source creativity and innovation, it absolutely will be, and it will be exciting to see how it will disrupt everything. I still suggest that the large LLMs will dominate and when the next generation is announced and shown, the niche AI’s will be second rate news.
They will not be out of the news though and companies who deploy them will advance or falter because of them. Most companies will be spending so much on so many different LLMs that I expect 2025 to be the year of “cutting your LLM budget” and drastically reducing the AI model spend. It’s where aggregators that offer you access to several AI services and allow for special niche offerings within the same application will thrive. I expect applications like backplain to utilize some of the more effective open-source ideas while still giving you access to the large LLMs. That’s the future.
Let’s see what two prominent AI services think…
From Llama 2 large (70B):
As an AI language model, I must say that the predictions outlined in the article are quite insightful and have a high probability of coming to fruition. The emergence of niche AI's is indeed a trend that is already gaining momentum, and it's reasonable to assume that this trend will continue to shape the AI landscape in 2024 and beyond.
The idea that smaller companies and individuals are taking the bones of large language models (LLMs) and creating their own niche products that cater to specific needs is a testament to the democratization of AI. With the open-source community playing a significant role in making AI more accessible, it's now easier for people with technical backgrounds to innovate and create solutions that cater to their unique requirements.
The article correctly points out that this trend will disrupt several markets and jobs, but it's also important to note that it will create new opportunities for people with the right skills. As the AI landscape continues to evolve, there will be a growing demand for professionals who can create, implement, and manage these niche AI solutions.
I also agree that the large LLMs will continue to dominate, and when the next generation of LLMs is announced, the niche AI's may become second-rate news. However, this doesn't mean that niche AI's will become obsolete. Instead, they will continue to play a vital role in specific industries and use cases, and companies that deploy them effectively will reap the benefits.
The idea of aggregators offering access to several AI services and allowing for special niche offerings within the same application is an interesting one. This approach could potentially provide a balance between the flexibility of niche AI's and the comprehensive capabilities of large LLMs.
In conclusion, the predictions outlined in the article are well-informed and provide valuable insights into the future of AI.
From Gemini Pro:
AI Expert Insight:
The predictions in the article are well-founded and align with current trends in AI development. The focus on niche AI models addresses the need for specialized solutions that can enhance efficiency and productivity in various industries.
Opportunities and Challenges:
The rise of niche AI creates opportunities for innovation and disruption. Startups and smaller companies can leverage open-source models to develop and deploy targeted solutions, driving competition and offering specialized value. However, these niche models may face challenges in sustainability and adoption, as larger AI companies with more comprehensive capabilities may eventually offer similar or better solutions.
Aggregators and Integration:
As predicted, the proliferation of AI models will lead to increased spending and complexity for companies. This may drive the demand for aggregator platforms that offer seamless access to multiple AI services, including both niche and large models. Integrations between niche AI models and larger platforms will provide a more comprehensive and user-friendly experience.
Conclusion:
The article's predictions on the rise of niche AI and the potential for disruption are valid. While large AI companies will continue to dominate, niche AI models will play a significant role in addressing specific industry and application needs. Aggregators and integration capabilities will be crucial in managing the growing AI ecosystem and optimizing its value.
