Listen to the audio version of the article
Google, Microsoft, Meta, probably Apple. And then the large producers of enterprise-class software such as Sas, Sap and Oracle. Or networking giants like Cisco. Big Tech has entered (or will enter) the arena of generative artificial intelligence and Amazon obviously couldn’t be missing from the call, both in terms of personal assistants for smart home devices (see the integration of LLM models in Alexa) both on that of infrastructures for cloud services and development environments. The latest announcement by AWS (which in the meantime has obtained the QI2/QC2 qualification from the National Agency for Cybersecurity to host critical data and workloads of the Italian PA) goes precisely in this direction and is, as it reads in the note released by the North American company, about five new Gen AI solutions that will make it easier to access high-performance “foundational models” (machine learning models pre-trained on large quantities of data such as GPT-3 and BERT), regardless of the reference sector. Amazon’s goal? Very explicit, and at the same time very ambitious: to unlock – for the benefit of companies of any size and sector (including startups) the full potential of generative coding to create differentiated apps, simplifying development processes and, at the same time, preserving privacy and safety.
Interoperability with Meta’s new LLM model
Among the innovations announced, the most important is Bedrock, a new service, announced in April and now officially released to the public, which makes the basic models of the main AI companies available through a single application programming interface (a single bees). Companies, developers and data analysts, in other words, will have more tools available to write and scale generative artificial intelligence applications useful for creating text, images, audio and synthetic data starting from a specific prompt. Furthermore, Bedrock is the first fully managed Gen AI service capable of offering Llama 2, Meta’s new generation large format language, guaranteeing user companies the necessary flexibility in using the models that best meet their specific needs .
Rethink the way you work
«Over the last twelve months – this is the basic consideration of Swami Sivasubramanian, vice president of the Data and AI division of AWS – the proliferation of data, access to scalable calculations and advances in machine learning have led to a surge in interest in generative AI, giving life to solutions and ideas that could transform entire sectors and rethink the way we work.” Among the new features made official at Amazon in these hours, some new features also stand out for CodeWhisperer (complementary A-based coding tool that generates full-line code suggestions) and QuickSight (server less business intelligence service native to the cloud), with the aim of facilitating the creation of new generative artificial intelligence applications and improving employee productivity. Finally, Embedding Amazon Titan is a large language model (LLM) capable of helping companies get the maximum power of any foundation model using proprietary data.