If you’ve spent any time researching artificial intelligence (AI) applications for your business, you might be wondering how this technology will change future interactions with your customers and even your employees. If you’ve tested ChatGPT you’ve probably seen that generative AI responses can be both fast and thorough. As with all technologies, there are limitations though--for example, the tool’s responses can be misleading and can also include bias. So what does the future potential like and what does this mean for your business?
Major Tech Companies Driving Innovation
The answer may come from one of the leading players in the tech industry. In January of 2023, Microsoft announced a multi-billion dollar investment in OpenAI (the developers of ChatGPT) to promote research and “safe, useful and powerful” development of AI. Through Microsoft’s Azure platform, they will expand access to large, advanced AI models. Businesses can work with Microsoft to apply for access to the most advanced AI models including GPT-3.5, Codex, and DALL-E 2.
Google is also on a mission to advance AI and bring its benefits to "everyone" through its Bard experiment. You have probably already experienced Google’s use of large language models (LLMs) in its products, such as Gmail’s ability to autocomplete sentences. Through Bard, they are building an interface to a LLM that will enable users to interact with its AI model. According to Google, they plan to advance the tool thoughtfully with input from industry experts, educators, policymakers, civil rights and human rights leaders, content creators and more.
Most recently Amazon and Anthropic have announced a strategic collaboration in AI, with Anthropic selecting Amazon Web Services (AWS) as its primary cloud provider. They will use AWS specific chips for building and deploying AI models. This partnership aims to make these models accessible to AWS customers and provide them with early access to customization features through Amazon Bedrock. AWS will become Anthropic's primary cloud provider, and Amazon will invest in Anthropic. The collaboration aims to leverage Anthropic's AI expertise with AWS's cloud technology for customer benefit.
Public vs Private Models
Most of these technologies are available to the public, supporting productivity and creativity use cases. Businesses have the opportunity to tap into off-the-shelf models that have been trained on public data (e.g. Bard, ChatGPT), develop proprietary models with in-house data, or use a hybrid approach.
If your business is interested in exploring AI, you will need to determine whether a public or proprietary AI solution will support the results you’re looking for. Here are some considerations:
What use cases are you looking for generative AI technology to solve?
By defining the use cases, you can determine whether you can rely on publicly available data, or need a build from exclusively in-house data sources. For example, if you’re using generative AI to generate new ideas or assist with copywriting, a public tool (e.g. ChatGPT) could work. Alternatively, if you need an AI tool to synthesize business results, you will need a proprietary model that has been trained on your internal sales data.
What level of control and/or privacy is required?
Concerns have been raised about public AI models and the potential for bias, misinformation, and copyright infringement. A private (open-source based or proprietary) AI model allows for more control and security regarding the data involved as well as the model’s outputs.
What resources and budget do you have available?
Owning your own Large Language Model (LLM) offers many benefits such as control, privacy, performance, and cost advantages. Training from scratch can be costly, but thanks to open-source foundational models is not a must anymore. Optimisation techniques make fine-tuning affordable and yield results surpassing GPT-4 on certain tasks. While serving an LLM can be challenging, leveraging ML platforms can streamline the process and make it affordable.
What about transparency and support?
With open-source LLMs, you have full visibility into the model’s inner workings, which can help build trust with your customers. Also, open-source communities are known to be fast moving and supportive, for constant and dedicated support help from a vendor is required.
The approach is nuanced and there is no one-size-fits all answer. The costs and recommended approach (e.g. public vs private model) will depend on your business objectives and desired outputs. Working with an AI consultant or technology firm, such as DataLux, may ultimately provide you with the right level of guidance and flexibility when considering your options. If you have questions on how generative AI can be applied to your business, please feel free to get in touch.