Uncategorized

AWS’s Response to Generative AI in 2023 (and What IT Pros Should Expect in 2024)


When 2023 began, Rada Stanic, chief technologist for AWS in Australia and New Zealand, imagined the year would be dominated by helping customers build more modern data platforms in the cloud, with a bit of artificial intelligence and machine learning thrown into the mix. That plan changed fast with the emergence of ChatGPT and a new era of generative AI.

“When generative AI happened, it took over nearly every single conversation we were having with our customers from different industries and really dominated the year,” Stanic told TechRepublic.

Stanic and Louise Stigwood, AWS director of enterprise in Australia and New Zealand, expect generative AI use cases among Australia and New Zealand-based customers to graduate to production grade in 2024. Enterprises in Australasia will also focus on integrating responsible AI and building data platforms that ensure AI initiatives are a success.

Jump to:

AWS reinvented its 2023 agenda as generative AI LLMs emerged

The overnight mainstreaming of generative AI caused AWS, like other cloud providers, to ramp up AI-related innovation in 2023, from strategy right through execution. This culminated in a number of big announcements at the hyperscaler’s November 2023 re:Invent conference.

Portarait of Rada Stanic, AWS chief technologist in Australia and New Zealand.
Rada Stanic, AWS chief technologist in Australia and New Zealand

SEE: Australia needs to prepare to reap the benefits of artificial intelligence.

These included a new generative AI-powered assistant, Amazon Q, targeted at businesses. AWS also expanded foundation model choice, fine tuning capabilities and responsible AI tools available within its new generative AI platform Amazon Bedrock, among other announcements.

“We have this plan of helping customers with everything from creating generative AI-powered services to providing the tools that help democratise the technology for developers and customers who don’t have strong technology skills,” Stanic said.

This is in line with efforts AWS has made with other products and services, including Amazon SageMaker, which launched in 2017 as a platform that makes it easier to deploy machine learning models in the cloud and is now used by 100,000 customers.

Regional AWS customers adapt to a new generative AI world

Customers of AWS also rode the generative AI wave in 2023. Local generative AI start-up Leonardo.Ai, which enables users to generate images and train their own custom models, generated 700 million images in the year to November 2023 — or 4.5 million images daily (Figure A).

Dashboard of generative AI platform Leonardo.Ai.
Figure A: With AWS cloud services, generative AI platform Leonardo.Ai is managing explosive growth in the volume of images its users are generating. Image: Leonardo.Ai

Scaling using Amazon’s Elastic Compute Cloud Inf2 instances for generative AI inference and for running several large language models, Leonardo.Ai’s was also the first Asia-Pacific team to complete AWS’ generative AI accelerator program to support its growth.

New Zealand-based telco One New Zealand, meanwhile, began using generative AI capabilities from AWS to help its contact centre agents better understand why customers call and to proactively support call resolution, leading to a measured 10% increase in customer trust.

Portrait of Louise Stigwood, AWS director of enterprise in Australia and New Zealand.
Louise Stigwood, AWS director of enterprise in Australia and New Zealand

Australian CEOs and boards look at AI and cloud cost optimisation

Louise Stigwood said generative AI became a topic for nearly every local CEO and board member in 2023 as they looked to understand the complexities of the technology and the role it may play in change. This led AWS to develop training courses for executives to get up to speed.

Cost optimisation remained another priority in a cost-constrained environment. Stigwood said organisations worked with AWS to double down on the modernisation of cloud environments to drive costs down, which saved millions over the course of last year for customers like NAB.

Many business customers also leveraged their savings to fund innovation. One example was the AFL’s announcement that Marvel Stadium would become the first stadium in the Southern Hemisphere to introduce Just Walk Out technology for checkout-free purchases at games.

The AWS technologies IT professionals should watch in 2024

AWS advances in 2023 are likely to pay dividends for customers this year. For example, the next generation two chip families — AWS Graviton4 and AWS Trainium2 — will support cloud workloads including ML and generative AI, helping to reduce cost and boost performance.

PREMIUM: Explore the evolution, features, drawbacks and benefits of generative AI.

Stigwood said new tooling for its AWS Supply Chain offering will help customers track and improve supply chain sustainability. Upgrades include the ability to request, collect and audit sustainability data from suppliers, including Scope 1, 2 and 3 carbon emissions data.

There are a number of other AWS improvements in 2023 that are relevant for local enterprises in 2024.

The emergence of artificial organisational experts

The synthesis of organisational information has emerged as a key use case for generative AI in general. Stanic said Amazon Q, Amazon’s new AI model built to help businesses leverage their proprietary data for AI securely, could have ground-breaking effects as usage grows this year (Figure B).

Picture of Generative AI virtual assistant Amazon Q unveiled by AWS CEO Adam Selipsky in 2023.
Figure B: Generative AI virtual assistant Amazon Q was unveiled by AWS CEO Adam Selipsky in 2023. Image: AWS

“It will act like an expert on a customer’s business,” Stanic said. “It has strong integration already with existing technologies like Amazon Connect for call centres. There is a strong interest in using generative AI assistants to help call centre agents in real time provide real help to customers.”

The democratisation of generative AI to build applications

Upgrades to Amazon Bedrock could make it easier for developers who are not AI experts to create generative AI applications. With 10,000 customers, Stanic said many enterprises are now in the proof of concept stage with applications and will move into production in the coming months.

For example, Australian customer Adore Beauty has used Bedrock to support the analysis of customer reviews to determine which products are performing well. Accounting software firm Xero created a virtual assistant to help onboard customers as well as answer their questions with AI.

Making responsible AI a part of organisational innovation

Discussions with customers about how to solve a problem or do something faster and better with AI often turn to questions of security and responsible AI, Stanic said. AWS is looking to integrate AI safety within its tools, as it has done with Guardrails for Amazon Bedrock.

Guardrails can help companies evaluate user inputs and foundation model responses based on use case-specific policies, providing an additional layer of responsible AI safeguards regardless of the foundation model. Stanic said it can help customers implement and evaluate models.

The take-off of a new network of low orbit satellites

Amazon’s Low Earth Orbit satellite project, which recently saw the company sign a deal with NTT Docomo in Japan to provide redundancy for its communications networks, could help telcos extend 5G capabilities or facilitate AWS to AWS communications off the public internet.

According to Stigwood, this could see AWS working with telcos in Australia and New Zealand to extend communications networks. Organisations like miners and others in regional areas could also benefit from the advantages of more connectivity options via Amazon’s LEO satellites.

AWS is expecting generative AI to take centre stage again in 2024. Other predictions for the Australian and New Zealand markets from Stanic and Stigwood include that local organisations will focus on data platforms, cloud migration and modernisation, and industry diversity.

The emergence of production-grade generative AI

While 2023 was dominated by understanding and experimenting with generative AI models, Stanic expects 2024 to be the year of production-grade implementation for generative AI in the region, which could make a big difference to a range of business verticals in the local market.

SEE: Australia has been embracing aspects of generative AI throughout 2023 while trying to stay ahead of the risks.

Stanic said employees in Australia and New Zealand are likely to be better able to leverage internal knowledge bases to find information quicker and easier, while the lives of developers will be changed for the better through coding companions like AWS’s CodeWhisperer (Figure C).

Screenshot of AWS’ CodeWhisperer that supports developers with productivity by generating code suggestions ranging from snippets to full functions in real time.
Figure C: AWS’ CodeWhisperer supports developers with productivity by generating code suggestions ranging from snippets to full functions in real time. Image: AWS

Stanic added that Australian enterprises will move towards transforming their customer experiences with generative AI, including through the popular virtual assistant use case for call centres.

“Those quite viable use cases we can expect to gain more traction this year,” Stanic said.

A focus on robust data platforms in the cloud

The success of generative AI will depend heavily on data. Stanic said organisations should not underestimate the impact of the underlying data and data platform on AI projects, because this can end up making or breaking the success of generative AI implementation.

“We think we will see customers doubling down this year on creating strong, scalable and high-performing data platforms in the cloud,” Stanic said.

AWS is removing some of the heavy lifting through Zero-ETL, which facilitates point to point data movement without ETL data pipelines.

“This can reduce data processing time and make the lives of data engineers’ lives much easier,” Stanic said. “It supports the critical role of data in powering success with generative AI.”

Continued cloud migration and modernisation

Migration and modernisation in the cloud is a trend Stanic said will continue in Australia and New Zealand, regardless of other trends that may come and go. She said organisations want to be able to operate more cost-effectively and scalably, and the cloud enables that for them.

This will be combined with cloud cost optimisation, after cloud costs have ballooned in recent years. With more controls available for cost optimisation and through support for making it iterative, AWS is working with customers to redirect cloud cost savings back into innovative projects.

New opportunities for tech industry diversity

The advent of coding companions like CodeWhisperer could begin to change how the market thinks about traditional technologies and software engineering roles, according to Stigwood. This could act as a catalyst for more diversity in Australasia’s technology industry.

“When there is diversity at the table, there are better outcomes, and that’s increasingly a focus, particularly for larger enterprises,” Stanic said. “This year, many have taken advantage of our program Cloud Up For Her, which helps females from any background become technologically fluent.”



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *