Artificial intelligence, this is how Amazon will challenge OpenAi and Alphabet
In the great race for the best artificial intelligence, Amazon challenges OpenAi and Alphabet by setting up a task force led by Rohit Prasad, former "Alexa man"
If the 4 billion invested in Anthropic , a young but promising AI startup by Italian-Americans Dario and Daniela Amodei , who "grew up" professionally in Microsoft's much more famous OpenAi before deciding to set up their own business, were not enough, Amazon seems intent on exploring the topic of 'artificial intelligence.
AMAZON WANTS TO ENTER THE OLYMPUS OF ARTIFICIAL INTELLIGENCE
The e-commerce giant led by Andy Jassy intends to develop a proprietary Large language model so as to officially follow the path already traced by the aforementioned OpenAI and Alphabet, the two most active US companies. The LLM on which Amazon is working, from the first rumors emerging from Reuters , is called " Olympus " and would already be able to count 2 trillion parameters.
ROHIT PRASAD'S TASK FORCE
Information on the Large language model cooked up by the e-commerce giant is still rarefied for the moment, but it is known that the task force led by Rohit Prasad , former head of the Alexa division, who will report directly to the CEO of Amazon, is working on it , Andy Jassy.
For the development of Olympus, Prasad involved researchers drawn from the Alexa AI developers and Amazon's scientific team to form the LLM. According to what we learn from Reuters , Jassy's aim, with the development of this new team and the related Large language model under development, is to internally create a model to make its offerings more attractive on AWS, where enterprise customers demand better performance. Furthermore, given the many connections with Alexa, nothing excludes the possibility of an enhancement of the virtual assistant on the horizon. What is certain is that, as highlighted by Reuters , Amazon is gradually increasing investments in LLM and generative artificial intelligence.
AMAZON'S INVESTMENTS IN AI
Amazon has trained smaller models like Titan in the past and has partnered with AI modeling startups like AI21 Labs. The latest intervention in this sense concerned the maxi financing for the aforementioned Anthropic, on which Amazon Web Services is betting four billion, the largest corporate deal made by AWS so far.
Under the agreements, Anthropic will move most of its software to Amazon Web Services' data centers and use the cloud computing company's internal chips to train models that power chatbots and other applications. Amazon engineers will have access to models from Anthropic, which had already made “Claude” available as part of Amazon’s “Bedrock” service.
This is a machine translation from Italian language of a post published on Start Magazine at the URL https://www.startmag.it/innovazione/intelligenza-artificiale-ecco-come-amazon-sfidera-openai-e-alphabet/ on Thu, 09 Nov 2023 12:40:23 +0000.