To Power Alexa, Amazon Is Developing An Improved LLM

(CTN News) – During Amazon’s first-quarter earnings call yesterday, Amazon CEO Andy Jassy said the company is building a “generalized and capable” large language model (LLM).

An LLM is a deep learning algorithm that can recognize, summarize, and generate text and other content based on knowledge derived from enormous amounts of text.

While Amazon has been using an LLM to power Alexa, the tech giant is currently developing one that is more capable than the current one.

The Amazon executive believes that the addition of an improved LLM will help the company develop “the world’s best personal assistant,” but acknowledged that it will be difficult across a variety of domains.

When people ask us about Alexa, we often explain that the investment would be much smaller if we were only building a smart speaker.

“However, we have a vision, which we hold in high regard, that we intend to build the world’s best personal assistant.

We will have a difficult time accomplishing this. There is a broad surface area that encompasses many different domains to consider.

Nonetheless, the advent of Large Language Models and generative AI makes the underlying models more effective, and I believe that this allows us to create the world’s most effective personal assistant.

With Alexa, Jassy believes that Amazon has a good starting point, since it has “a couple hundred million endpoints” spanning the entertainment, shopping, and smart home sectors. Additionally, third-party ecosystem partners are heavily involved in the project.

A much larger, more generalized, and capable language model is being developed in addition to the existing one. As a result, we will be able to become the world’s best personal assistant very quickly. The business model behind it is significant.”

Jassy indicated that Amazon has invested heavily in AI and LLMs for years, but smaller companies cannot do the same, which is why the company launched Bedrock earlier this month.

Using pre-trained models from startups such as AI21 Labs, Anthropic and Stability AI, Bedrock allows you to build generative AI apps.

Furthermore, Bedrock offers Titan FMs (foundation models), a family of models that are trained in-house by Amazon Web Services.

It has become increasingly popular since ChatGPT launched last year. ChatGPT is generating so much buzz that major tech companies are looking for ways to incorporate LLM-based improvements into their own products to stay competitive in the fast-paced AI space.

The Information reported yesterday that Apple is developing Siri improvements using LLM. In addition, Google’s Assistant is likely to do something similar.

The Amazon quarterly call wasn’t the only way Alphabet, Microsoft, and Meta discussed their investments in large language models.

A CEO at Alphabet said the company would continue to incorporate artificial intelligence into search, while a CEO at Microsoft, Satya Nadella, noted that the company has already seen a significant increase in use of Bing since its ChatGPT integration was introduced.

Moreover, Meta CEO Mark Zuckerberg said the company will invest in artificial intelligence and introduce AI-related updates.

Amazon reported first-quarter earnings that beat expectations and initially sent shares soaring, but later reversed course after executives raised concerns about cloud growth.

In the third quarter, revenue increased 9.4% to $127.4 billion, while operating income was $4.8 billion.

SEE ALSO:

Shares Of First Republic Are Down 40%, Sources Say; FDIC Receivership Most Likely

With Bitcoin’s Price Rising, MicroStrategy Is In a Unique Position To Benefit: Berenberg