In a bid to break free from OpenAI, companies are building their own custom AI chatbots
OpenAI has long dominated the generative AI market, with its GPT-4 model being hailed as the best-performing model to date. However, an increasing number of businesses are now opting to develop their own smaller AI models that are tailored to their specific needs. This shift highlights the desire for businesses to have more control and customization over their AI chatbot solutions.
One example of this trend is seen with Salesforce, which has recently piloted two coding AI assistants called Einstein for Developers and Einstein for Flow. These assistants are trained not only on Salesforce’s own programming data but also on open-source data. The aim is to create small AI models that are specifically designed for niche business applications. Although these assistants are capable of tasks like writing poems, they may not excel in areas where they haven’t been trained on the broader internet, unlike OpenAI’s ChatGPT.
While OpenAI, Google, Amazon, and Meta are focused on developing larger and more powerful AI models, there is still a strong case for companies to explore their own AI solutions. As a result, we may likely see an influx of smaller AI models specifically designed for particular tasks. This implies that people may interact with different AI bots throughout their day, depending on the activity at hand. Yoon Kim, an assistant professor at the Massachusetts Institute of Technology, highlights that companies can adopt AI in a more cost-effective manner by focusing on specific applications.
Braden Hancock, the chief technology officer of Snorkel AI, has been assisting businesses, particularly those in the financial sector, in building small, specialized AI models that power task-specific bots. These bots are designed to perform a single function, whether it’s customer service assistance or coding support. Hancock explains that the initial worry among businesses about the all-encompassing capabilities of AI, particularly following the release of ChatGPT, proved to be unfounded. In reality, there are few business applications that can be fully addressed by an out-of-the-box AI model without any modifications.
The future of AI development hinges on a couple of potential scenarios. If hardware costs decrease significantly, GPT-4 or similar models may become the go-to solution for businesses, catering to all their needs. However, another scenario could arise where the market is flooded with more large-language models (LLMs), creating increased competition for OpenAI. This could explain why OpenAI has been advocating for additional regulation to establish an advantage over its AI competitors and make it more challenging for others to enter the market.
The emergence of tailored AI models presents a new landscape for businesses, giving them the opportunity to create bespoke solutions that align with their specific requirements. However, it remains to be seen whether these smaller models will revolutionize the AI industry or if larger models will continue to dominate. The ongoing advancements in AI technology, along with the introduction of more cost-effective hardware, will certainly play a pivotal role in shaping the future of AI chatbots and their integration into various industries.
In a world full of rapidly evolving AI technology, businesses must carefully consider their options and weigh the advantages and limitations of both pre-existing AI models and their own custom-built solutions. The decision ultimately rests on finding the optimal balance between tailoring AI to meet specific business needs and benefiting from the continuous advancements in AI capabilities.
