After the popularity of generative AI, Amazon Cloud Technology: Cannot rely on a single large model

Tech 2023-07-10 10:41:33 Source: Network

Global Network Technology Reporter Li Wenyao: Since the global outbreak of generative AI, technology industry giants have launched AI big model products. The industry generally believes that with the emergence of generative AI, future industry development will be rewritten, and its importance is no less than an "industrial revolution"

Global Network Technology Reporter Li Wenyao: Since the global outbreak of generative AI, technology industry giants have launched AI big model products. The industry generally believes that with the emergence of generative AI, future industry development will be rewritten, and its importance is no less than an "industrial revolution".

Amazon Cloud Technology clearly does not want to miss the development trend. At the China Summit held in June, Amazon Cloud Technology revealed that it is building out of the box generative AI applications for some common scenarios, further lowering the threshold for using generative AI.

Dr. Matt Wood, Vice President of Global Products at Amazon Cloud Technology, also made his first appearance at the China Summit after the pandemic. When discussing the development of generative AI big models, he told reporters that a single big model is not omnipotent, We believe that we cannot rely on a universal, single large language model to handle various tasks. We believe that the correct approach should be for customers to access multiple models and customize their own models based on their own needs and data

Cannot rely on a single large model

In April of this year, Amazon Cloud Technology released a tool called Amazon Bedlock globally to help businesses develop generative AI services. Companies can customize basic generative AI models and create unique proprietary models using their own data.

It is worth noting that in order to better improve the efficiency of large models and reduce costs, Amazon Cloud Technology has also announced the official availability of Amazon EC2Trn1n and Amazon EC2Inf2 instances, providing the most cost-effective generative AI cloud infrastructure.

In addition to Amazon Cloud Technology, technology giants such as Google and Microsoft have already laid out generative AI tool applications. In the domestic market, companies such as Baidu, Alibaba, and Tencent have launched related products. In some fields such as office, healthcare, and search, vertical generative AI applications are also constantly landing.

This will drive the development of the global economy. According to a report released by global consulting giant McKinsey&Company, the introduction of generative artificial intelligence (AI) tools is expected to contribute trillions of dollars in added value to the global economy annually. The report points out that this increase is basically equivalent to the GDP of the UK in one year ($3.1 trillion in 2021).

In MattWood's view, after the rapid development of generative AI, the industry needs to think about what systems need to be built, what resources are needed, and especially what large-scale models are needed to achieve the overall technical and application vision.

As a provider of technical solutions, this is not an easy task. The needs of different industries, enterprises, and development stages have their own characteristics.

Therefore, in this process, Amazon Cloud Technology first needs to ensure the universal availability of generative AI and deliver services to users at extremely low cost and with the lowest possible delay. We do not believe that there is a so-called universal big model that is controlled by others. We believe that it should be built on top of the big model to ensure that users can customize their own models in a private and secure manner, "said MattWood.

More and more third-party models will emerge

The common problem with universal models is that they are widely applicable to various types of problems, but their depth is insufficient and their expertise in vertical applications in specific industries is still insufficient.

The generative AI model provided by third parties has become crucial. It is understood that currently, Amazon Cloud Technology has introduced partners including Antiopic, AI21Labs, and StabilityAI, as well as a pre trained Amazon Titan basic model developed by Amazon Cloud Technology.

With the help of the Amazon Titan series of basic models, customers can build specialized models based on their own needs and big language models, and use their own data. And these models can provide more in-depth industry solutions.

In the past period of time, we have seen clients construct models based on their own needs in various ways, "exclaimed Matt Wood.

Currently, many industry models have been trained on Amazon Cloud Technology platforms. For example, Bloomberg runs a very large financial services model on Amazon Cloud Technology. The same method is also applicable to other enterprises, for which they can choose the corresponding basic model based on their own industry and then train the model. In addition, enterprises can supplement and add unique private information based on their own situation, so that the output of the model is highly relevant to the problems or requirements they propose.

For the development of future collaboration models, MattWood believes, "We can foresee an increasing number of third-party models, including both our own third-party models and open source models. At the same time, there will also be more Amazon Titan models available for customers to choose from

GenerativeAIStill in its early stages

GenerativeAIAI6GenerativeAIAIGenerativeAI

Matt WoodGenerativeAI

GenerativeAIMatt WoodGenerativeAI

Therefore, Amazon Cloud Technology has made a lot of efforts in content screening and filtering, filtering and checking the input information and data, and also verifying the answers generated by the model to ensure that the responses provided are healthy and harmless.

Matt WoodGenerativeAIMatt WoodGenerativeAI

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])

Mobile advertising space rental

Tag: After the popularity of generative AI Amazon Cloud Technology