Tech Cloud reports original.
If this is the era of AI big models, it is better to say the era of generative AI.
Among the three types of AI big models, generative AI, and ChatGPT, generative AI is the broadest concept, covering all applications that use AI to generate new content.
Large models are a way to implement generative AI, while ChatGPT is a specific application of large models and generative AI in practice.
When the performance of ChatGPT shocked the world, domestic and foreign technology giants set their sights on the large model used behind ChatGPT and practiced it.
Unlike tech giants that take the big model route, Amazon Web Services has set its sights on generative AI from the beginning.
Recently, at the 2023 Amazon Web Services China Summit, Zhang Wenyi, vice president of Amazon Global and executive director of Amazon Web Services Greater China, said that big models are not the whole of generative AI, but an underlying basic platform of the generative AI ecosystem.
Wenyi Zhang, Vice President of Amazon and Executive Director of Amazon Web Services Greater China
"The mission of the platform is to make it easier for people to build machine learning applications on top of it, and use generative AI to solve problems in their own specific fields or industry scenarios, which is the key factor to truly change the industry in the To B field," said Zhang Wenyi.
The opportunity presented by generative AI
Generative AI is an AI that uses machine learning techniques, especially deep learning, to generate previously unseen content, including but not limited to the following types:
Generative Adversarial Networks (GANs): This is a special deep neural network structure consisting of two subnetworks, one as a generator and the other as a discriminator.
The generator tries to generate fake data, and the discriminator tries to tell if the data is real or not. Through this adversarial process, the generator is able to learn to generate increasingly realistic data.
Variational Autoencoders (VAEs): This is a generative model that uses a neural network to encode input data, such as an image, into a latent space, and then decodes new, raw data from that space.
VAEs are a great way to learn the intrinsic structure of data and generate new data.
Autoregressive model: This model predicts the next data point based on the previous data point and can be used to generate a continuous series of data, such as text, music, and so on.
Transformer Models: Such as OpenAI's GPT-3, which uses self-attention mechanisms and transformer architectures to generate text, is a very important class of models in modern natural language processing.
These generative AI models have a wide range of uses in a variety of applications, including art generation (such as DeepArt or DeepDream), text generation (such as chatbots and news generators), music generation, and more complex tasks such as video generation and virtual reality environment generation.
Specifically, in the gaming industry, generative AI can create new game environments and characters, resulting in richer gaming experiences. For example, AI can generate endless maps, making game worlds larger and more diverse.
In addition, the life sciences industry is also expected to achieve breakthroughs due to advances in generative AI. Generative AI can help scientists discover disease-related proteins faster, further accelerating the drug discovery process.
Not only that, special basic models are used to predict molecular properties, guide the generation of new proteins with better physiological functions, and discover new drugs to treat diseases.
With the help of generative AI, more currently incurable diseases are expected to be cured, and human lifespans will be further extended.
In the manufacturing industry, generative AI can automatically generate new designs through existing design training models, and designers can upgrade and improve on the basis of these designs, thereby reducing the labor cost of design and improving output efficiency.
In the production line, every robotic arm, every device, and every production line in the future will have the opportunity to control through generative AI, improve the level of automation, and further promote the intelligent development of the manufacturing industry. Data shows that by 2027, 30% of manufacturers will use generative AI to improve the efficiency of product development.
In addition, generative AI has a lot of room for imagination in education, social and public welfare, and creative industries. As technology continues to evolve, it may lead to more opportunities in the future that we can't imagine today.
Goldman Sachs predicts that generative AI could boost global GDP by 7% to nearly $7 trillion. McKinsey also estimates that this technology can increase the productivity of 2,100 specific jobs in 850 occupations around the world, and reduce production costs by as much as $6.1 to $7.9 trillion.
What we can sense is that generative AI is reshaping our lives, and the scale, depth and breadth of the changes it could trigger may only be matched by the explosion of the internet.
How to unlock the value of generative AI?
Generative AI is the torrent of scientific and technological reform, which brings unprecedented historical opportunities, provides unlimited imagination space for enterprises, and makes the research and development process of products and services more efficient and interesting.
But specific to application practice, what should enterprises do in this torrent of the times? As the world's leading cloud service provider, Amazon Web Services has given its own solution ideas:
First of all, help enterprises find the right scenario and the basic model of this scenario, help enterprises use their own private data combined with the basic model to build their own customized model "easier", and at the same time ensure that their "private data security" will not be absorbed by the basic model.
In this regard, Amazon Web Services launched Amazon Bedrock, enterprises can flexibly choose the model that suits them, build applications more easily, and develop customized models on the basis of ensuring data security and privacy, without the need for a large amount of annotated data.
Second, it helps enterprises obtain out-of-the-box generative AI applications in some general scenarios, further reducing the threshold for use.
Amazon Web Services has also launched an AI programming assistant, Amazon CodeWhisperer, which can generate code suggestions in real time according to developers' natural language instructions through the embedded basic model, greatly reducing developers' heavy work.
Third, with the universalization of enterprise generative AI applications and the continuous iteration of basic models, there must be ultra-large-scale and cost-effective cloud platforms to support continuous model training and application-side large-scale inference.
Amazon Web Services offers a full range of compute, high-speed networking, and high-performance storage options. In addition to the common CPU and GPU options in the industry, Amazon Web Services has more than 5 years of experience in self-developed chips.
In addition, for the problem of talent, Amazon Web Services has also given a solution.
Amazon Web Services also has a wealth of professional technical support resources, including SA, product experts, artificial intelligence labs, data labs, rapid prototyping teams, and professional service teams to help customers solve the engineering challenges of the last three kilometers of application generative AI.
At this summit, Amazon Web Services also launched the "Amazon Web Services Startup Accelerator" for startups, from basic model development, the creation of new consumer applications or industry applications, to toolchain optimization, to help more enterprises quickly embrace generative AI.
Cloud vendors, "underlying architects" in the era of generative AI
Cloud services are a key productivity to support digital innovation.
In fact, it is not difficult to find that behind every technological progress, cloud vendors play an important role.
Behind this wave of AI, we can see that cloud vendors have provided infrastructure, AI services and application tools for AI research and development, and have also played an active role in promoting AI research, development talent training and practical application.
This is the case with Amazon Web Services. In addition to the above AI services and application tools, Amazon Web Services also provides rich computing resources and storage services for the market.
In the face of the blowout computing power demand brought by the generative AI era, Amazon Web Services provides better cost performance through self-developed chips, responds to sudden computing power requirements through a variety of rich combinations of computing, network, storage and other products, and effectively reduces the complexity of operation and maintenance through Serverless, thereby simplifying the use of computing power and fully meeting the diversified computing power needs of users.
"For the global layout, Amazon Web Services provides solutions for a variety of products from the center to the edge, including global infrastructure, the ability to quickly deploy stable systems, and the ability to fully support business compliance in various countries and regions around the world, which can become the cornerstone for users to create a solid underlying architecture." Xiaojian Chen, General Manager of Amazon Web Services Greater China Product Division, added.
In this way, it is not an exaggeration to compare cloud vendors to the "underlying architects" of the generative AI era.
Cloud vendors must not only play the role of "underlying architect" in the era of generative AI, but at the same time, cloud vendors must overcome challenges such as data security and privacy protection, provide users with safe and convenient services, and enable the application of generative AI to penetrate more widely and deeply into every industry and field.
This means that they need to continue to provide strong technical support, but also assume social responsibility to promote the healthy development of AI technology and help the popularization and application of AI.
Cloud vendors, as "underlying architects," need to play a role in several important ways. First, they need to continue to develop and optimize AI technologies, ensure their reliability and security, and provide users and developers with more powerful and easy-to-use tools.
Second, cloud vendors need to actively participate in the formulation of AI standards and specifications to guide the healthy development of AI technology. In addition, they also need to take on the important task of cultivating AI talents and promoting the popularization and application of AI.
In the future, we expect cloud vendors to continue to play their role as "underlying architects" to lead the development of generative AI technology and help the whole society realize the huge potential of AI.
【About Technology Cloud Report】
An enterprise-grade content expert focused on originality – Tech Cloud Reporting. Founded in 2015, it is a top 10 media in the cutting-edge enterprise IT field. Authoritatively recognized by the Ministry of Industry and Information Technology, Trusted Cloud, one of the official designated communication media of the Global Cloud Computing Conference. In-depth original reports on cloud computing, big data, artificial intelligence, blockchain and other fields.