So Amazon Web Services wants to make artificial intelligence more democratic

According to McKinsey, generative artificial intelligence will lead to an increase in turnover in numerous production areas, which will range from a minimum of 2.6 trillion dollars to a maximum of 4.4 trillion dollars globally. But the risk is that this new industrial revolution could overwhelm small and medium-sized companies and favor the big names in technology. And so the idea of ​​AWS seems to open a way out: democratizing artificial intelligence is the common intention that binds the proposals of the American company. “We tried to understand why this technology is still so complicated, and we saw that it is very difficult to develop new models and integrate them into technologies, applications, websites and business projects,” says Julien Groues, General Manager, Italy and France, AWS. “We want to make artificial intelligence accessible: we started with Sagemaker, used by over 100,000 people, which allows application developers to implement AI in their projects, without having to be a data scientist. Now we have developed Bedrock, which allows you to easily access a package of libraries for all your needs”.

What are they?
“There are several challenges in applying AI in practice. First, developing models requires a lot of time and considerable computing power. We want to give access to all available models on our platform: we have some developed by Amazon such as Amazon Titan, but also Anthropic, which is a Foundation model, Stability.AI is great for working with images, Jurassic for translations. The offer evolves, because what was the best model five years ago is not always the same today, so we update them and add new ones constantly. The second point concerns a problem that you were the first to face in Italy: the use of artificial intelligence models and respect for corporate and personal data. This is crucial, because if AI is being used to schedule a meeting or launch a marketing campaign, it is essential that the data is not disclosed. For years, all our systems have been GDPR compliant, and as a result the use of AI on this data remains compliant and the data itself remains within the company undisclosed, including results and prompts. The third relevant aspect is cost management. Current technologies are still economically demanding, and to make them accessible to a wider audience they need to be not only simpler, but also cost less. That’s why we developed our own AI chips, which allow us to be independent of vendors like Nvidia. Not all AI processes need as much processing power, and our two chips, Inferentia2 and Trainium, offer more performance at a lower cost.”

Where is the significant impact for a company?
“First of all, improve the customer experience. For example with intelligent agents and assistance robots. Not only chatbots, but also virtual assistants, who recognize the customer when he connects to the website or makes a call, and provides a personalized interaction. We have been offering this experience with voice for years with Alexa, for example. Secondly, it is essential for the productivity of the company to ensure that the employee or the company team can easily access the information. With Foundation Models the research becomes faster, you can work on the generation of emails, texts, video presentations and minutes of meetings.There are also tools like CodeWhisperer, very useful for application developers, who can now use AI to facilitate their work The third aspect concerns business operations, such as document management, content generation and continuous data creation. Finally, creativity. With AI it is possible to create new images, edit videos, compose music, completely transform a image, create animations. The AI ​​can also generate text for Twitter campaigns, turning those ideas into a video.”

Why would a business choose AWS if Microsoft built AI into Office and Adobe built it into Photoshop for image generation?
“First, these are limited systems using a single technology. We believe companies need access to a broad range of models and frameworks to be able to use the technology that best suits their needs, with the flexibility to adopt new approaches as better ones emerge. Secondly, we work on technologies that respect privacy; our privacy and security guidelines are among the best in the industry. Third, we offer an accessible system. Developers can easily use our technologies, switching from Sagemaker to Bedrock to build new applications. Furthermore, with Codewhisperer, anyone can use artificial intelligence to create applications simply by using a text prompt. Finally, there is our expertise. We have been using artificial intelligence to help companies for more than three decades”.

Shopping recommendations, Prime Video tips, Alexa experience — what has been most helpful in growing AWS AI?
“Everything: Speech recognition with Alexa has been very important, as have the technologies we implement in call centers and in Prime. Equally critical are data and image analysis in distribution centers and predictive maintenance, which has become one of the most important uses of AI for businesses. Alexa uses over 30 machine learning models and delivers more than a billion responses every week—that’s staggering volume, but it’s also the reason these interactions work so well. In addition, we have thousands of developers in Amazon and AWS working on these technologies, bringing their skills and experience”.

Artificial intelligence

Dialect, cooking and jokes: this is how Alexa learns Italian

by Bruno Ruffilli

August 19, 2021

But if we see the most sensational examples of generative artificial intelligence right now, the first name that comes to mind is ChatGPT, not Alexa.
“Too bad, it means that there are better companies than us in marketing”.

Are there any facilitations for startups or students?
“For startups, we offer specific programs to assist these emerging companies in the launch, test and use of our systems. AWS Activate is the program that allows startups to experiment with our technologies at no cost. Free online training programs are also available for students, but actually for everyone”.

Which companies are more inclined to adopt AI-based solutions?
“We have constant interactions with companies of all sectors and sizes. Undoubtedly, large companies are embracing AI faster, but smaller companies are also realizing that they can grow dramatically with these services. Increased interest from small and medium-sized enterprises. And there are many students and startups using our systems, such as advertising and marketing companies. Currently, it seems that everyone is considering leveraging AI in some way.”

And what about hallucinations, fake news, invented answers?
“Good question. It is essential that the last word in the decision-making and control process is entrusted to a human being”.

Artificial intelligence

10 AI-powered services you need to know and use today

by Andrea Nepori

07 April 2023