Amazon Web Services (AWS) recently unveiled its latest contribution to the AI ecosystem: Amazon Bedrock.
After an intense four months of development and testing, this suite of foundational AI models is now ready for use.
Amazon has gone down the infrastructural route in AI development, offering tools through AWS while roping in developers like Anthropic, with whom they plan to invest $4 billion.
In short, Bedrock is a managed service that enables users to build, tune, and deploy a host of foundational generative AI models.
By July, AWS added cutting-edge models to the platform, such as the Anthropic Claude 2, Stability AI SDXL 1.0, and Amazon’s own Titan model.
A critical feature of Amazon Bedrock is its adherence to regulatory standards, especially aligning with the European Union’s GDPR. Enterprises can use the platform to safely customize models with their own data while remaining compliant.
As AWS dives deeper into the generative AI space, it’s shaping up as a compelling alternative to established open-source platforms, and reports indicate that AWS is gearing up to integrate Meta’s advanced LLM Llama 2 into Bedrock.
The AWS advantage
The AWS advantage over open-source is offering models through a well-tuned API, integrated seamlessly within their infrastructure. This is bound to stir the pot with Hugging Face, the go-to host for these models for quite some time.
Despite AWS offering similar models to Hugging Face, whether the developer community will make the switch remains to be seen.
With that said, Amazon and Hugging Face, while competitors on some level, are also collaborating. They combined forces in February to make open-source AI more scalable through AWS.
Tech giants and AI developers seem to be re-drawing the battle lines. Amazon is throwing down the gauntlet to OpenAI, which has been slow to provide model access to hopeful developers. Moreover, the fluctuating performance and ‘drift’ exhibited by OpenAI models put many enterprises off moving their AI workloads.
However, OpenAI has leveled up its offerings of late with advanced image detection, and Google is allegedly preparing to announce a new model dubbed Gemini, though that remains speculative.
The generative AI industry seems to be maturing as we head towards 2024.