Amazon is looking to be a big player in the generative AI space and has launched Amazon Bedrock to offer access to pre-trained models from startups like AI21 Labs, Anthropic, and Stability AI.
Bedrock provides a way to build generative AI-powered apps and also includes access to Titan FMs, a family of models trained in-house by Amazon Web Services (AWS). AWS has been collaborating with startups in the generative AI space and has also launched a generative AI accelerator for startups.
Bedrock aims to be different from other AI model hosting services as it is aimed at large customers building “enterprise-scale” AI apps. This differs from some of the AI model hosting services, like Replicate, Google Cloud, and Azure.
It’s worth noting that no customer data is used to train the underlying models, and AWS customers can tap into AI models from various providers. It’s unclear what the terms of the model licensing or hosting agreements are and if the generative AI model vendors were incentivized by AWS to join Bedrock. According to estimates from Grand View Research, the generative AI market could be worth close to $110bn by 2030.
Bedrock also offers access to third-party models, including AI21 Labs’ Jurassic-2 family, which can generate Spanish, French, German, Portuguese, Italian, and Dutch text. In addition, Anthropic’s model, Claude, can perform a range of conversational and text-processing tasks. At the same time, Stability AI offers a set of text-to-image models called Bedrock-hosted models, which includes Stable Diffusion. These models can generate various types of visual content, such as images, artwork, logos, and graphic designs.
Another milestone achieved by Amazon in the AI arena is the development of CodeWhisperer, an Ai-powered code-generating service. Yes, you read that right! Moreover, it is free of charge without any usage limitations.
GitHub’s Copilot, its chief rival, already had over a million users in January, including thousands of enterprise customers. So, CodeWhisperer has some catching up to do, and they plan to do just that by launching the CodeWhisperer Professional Tier for corporate clients.
The CodeWhisperer Professional Tier comes with single sign-on and AWS Identity and Access Management integration. It also has higher limits on scanning for security vulnerabilities, making it perfect for businesses with more complex needs.
And now, CodeWhisperer supports even more programming languages – including Go, Rust, PHP, Ruby, Kotlin, C, C++, Shell scripting, SQL, and Scala.
The purpose of building the Titan models was to identify and eliminate any “harmful” content that may be present in the data provided by AWS customers for customization. The models were also designed to reject any “inappropriate” content that users may enter and filter out any output containing hate speech, profanity, or violent language.
The Titan FM family currently comprises two models, a text-generating model, and an embedding model. The text-generating model can write blog posts and emails, summarize documents and extract information from databases. In contrast, the embedding model translates text inputs into numerical representations containing the text’s semantic meaning.
Several lawsuits involving generative AI technology from companies such as OpenAI and Stability AI are pending. The plaintiffs claim that copyrighted art and other data were used without permission to train the generative models.
One of the cases aims to determine whether code-generating models that don’t provide attribution or credit can be commercialized. Additionally, an Australian mayor has threatened to sue OpenAI for defamation due to inaccuracies generated by its generative model ChatGPT.
Despite the potential legal and ethical issues surrounding generative AI, the market is booming, with companies like OpenAI and GPT-3 leading the charge. Amazon’s entry into the market with Bedrock is a significant move and could help AWS solidify its position as a leader in the AI space.
Overall, Bedrock is Amazon’s most forceful play yet for the generative AI market. While there are some unanswered questions around pricing and licensing, the platform’s potential to scale AI applications and create new business opportunities is significant. As with any emerging technology, however, it will be necessary for companies to approach generative AI with caution and consider the legal and ethical implications of using such models.
If you want more insightful stories like this then Hayvine is the way to go. Find interesting news on various topics and be the first one to know the latest update in town!