(Photo/Associated Press)

AI artificial intelligence technology has accelerated the pace of digital transformation for enterprises in recent years, and at the same time accelerated the expansion of innovation momentum for cloud applications. In response to the new trends that AI technology will bring from 2023 to 2025, Google Cloud released the latest forecast report. , in the next 2 to 3 years, related technologies and AI tools based on cloud applications will not only bring more time-saving and labor-saving work efficiency to enterprises, but also more breakthrough revolutions will appear, mainly covering ten aspects, including: The trend of AI expansion that helps realize the "three days off a week" is coming, and more than half of enterprise applications are created by non-IT developers.

The top ten trends of cloud application development predicted by Google in the next three years are summarized as follows:

Please read on...

1. AI solutions help promote the "three-day weekly rest" system

In the past few years, AI technology and related products have been proven to improve employee efficiency.

Taking the integrated machine learning platform Vertex AI as an example, the platform helps data scientists import machine learning models into the actual working environment faster by automating routine tasks such as model management, monitoring and version control.

Now Vertex AI can help build and train machine learning models 5 times faster, allowing data scientists to save time on experiments, reduce custom code work, and import more machine learning models into the actual working environment.

And Vertex AI Vision now provides a fully managed development environment suitable for creating computer vision applications, which can greatly reduce the time required to build and deploy related applications. What used to take weeks of work time can now be completed in just a few hours .

In addition, other Google Cloud AI tools also provide various assistance. For example, using Contact Center AI, the number of cases that the customer service center team can handle simultaneously has increased by 28%.

Likewise, sales and e-commerce teams saw a 40% increase in customer conversion rates with Recommendations AI.

More and more companies are choosing to complete their work in a smarter way, and the efficiency improvement brought about by the adoption of AI technology means that the vision of "three days off a week" is likely to come true.

(Photo provided by Google)

2. More than half of enterprise applications will be created by non-IT developers

With the development of democratization of technology, "low-code" or "no-code" technology is familiar to more companies, and it is expected that more and more development work will be undertaken by teams or personnel outside the IT department .

According to market research firm Gartner, by 2025, as many as 70% of emerging applications developed by enterprises will be created using "low-code" or "no-code" technologies.

Much higher than in 2020, only nearly 25% of applications are developed using this technology.

This means that people with non-IT technical backgrounds can also create enterprise-level applications that meet their needs, and they can create applications and automate workflows by themselves without having programming skills.

Developers still have to build these no-code tools and provide safeguards to keep businesses safe, but business users can do a lot more without the help of developers.

This transformation makes the operation of the enterprise no longer solely dependent on specific IT professionals, but also allows people at different levels to have the opportunity to participate in the development.

3. Obtain actionable real-time data through machine learning

How to make real-time "data-driven decisions" is more critical than simply obtaining information.

First, enterprises must be able to see and trust data; Google Cloud's Dataplex can now automatically classify all Google Cloud Platform (GCP) data owned by enterprises and integrate data from third-party sources.

In addition, in order to provide more reliable information, Dataplex's data quality and data lineage functions use smart technology and automation to ensure that data can produce accurate insights that can be used as a basis for action.

Dataflow, a stream analysis service currently used by Google services, allows enterprises to start small when using this service, and then expand to deal with real-time events, such as providing traffic information of a certain place when the application needs it.

Google Cloud predicts that other features will be launched in the next few months, so that more companies can make full use of existing data and continue to create business success.

It is estimated that by 2025, 90% of the data will be converted into actionable information in real time through machine learning technology.

4. Improve reliability through open source management

The speed, flexibility, and scalability that open source offers make developers highly dependent on the use of open source.

According to Market Guide for Software Composition Analysis (Market Guide for Software Composition Analysis) by market research firm Gartner, almost all enterprises will use open source software, but this has also become one of the pain points for enterprises to be attacked by hackers.

In order to avoid hacking attacks, enterprises have also begun to pay more attention to the importance of open source code management. In addition to maintaining the open nature of the source code, it can also improve software reliability.

In addition, open source "custodians" not only find security holes, but also help fix them.

Managers update old dependencies and keep track of newer dependencies.

In the case of open source management, dependencies have built-in automated testing mechanisms and may provide responsive service level agreements.

This is especially important today, as there are many digital infrastructures that rely on open source to function, and these source codes are maintained by the community in an "as is" manner.

This situation will change due to the popularity of open source management, such as Google Cloud Software Delivery Shield is a tool for managing open source.

This fully managed security solution protects the enterprise software supply chain from source code through deployment.

Google scans, analyzes and fuzzes over 250 Java and Python packages to find security holes for businesses and update them as needed.

Because open source management can greatly improve security, Google Clou expects all enterprise developers to adopt this practice by 2025.

5. Protective measures will move towards automatic code mining management

In the face of endless cyber attacks, it is difficult for today's security operation team to stop the harm.

According to a survey by Mandiant, a global information security company, intrusions involving ransomware in the Asia-Pacific region will increase from 12.5% ​​in 2020 to 38% in 2021, indicating that enterprises in the Asia-Pacific region are facing more and more frequent cyber security threats.

In order to implement protection measures at scale in the cloud environment, Google Cloud uses code to improve the agility and application scale of security mechanisms.

The process will employ an API-first approach, Chronicle and other tools, including Google Cloud's latest Mandiant web security portfolio.

Access rules for confidential information can be automatically established based on administrator activity records and other data, so there is no need to manually set and copy them to different environments.

Similarly, users can also create response playbooks to automate workflows such as capture, analysis, and response, allowing security analysts to work closer to developers, saving time to focus on major threats facing the enterprise.

Google Cloud predicts that by 2025, 90% of security measures will be automated and managed in the form of code.

6. Designing inclusive of neurodiversity increases user adoption

Whether it is studying, working or running a relationship, all kinds of people experience, interpret and deal with people and things in the world around them in different ways, and neuro-inclusive design (Neuro-inclusive design) takes into account the unique needs of different people , thereby increasing user adoption.

When designing interactive and visual features, consider the presentation of sound effects, vibrations, or pop-up windows, which can stimulate the senses and be distracting.

Google Meet's closed captioning is a good example of neurodiversity-inclusive design, a feature that allows users to more accurately understand information visually and allows attendees who speak different languages ​​to communicate effectively.

Developers are already incorporating neurodiversity-inclusive designs into Google products, which is expected to increase user adoption by a factor of five over the next 2 years.

7. Efficiency improvements in transactional and analytical workloads

In the case of data architectures in the past, transactional and analytical workloads were separated from mixed workloads because they used different databases.

Transactional databases are optimized for faster reads and writes, while analytical databases are optimized for aggregating large datasets.

In the case of personalized recommendations for an e-shopping site, the application must support both workloads with the same data set and avoid performance degradation.

Google Cloud now offers an integrated platform for both workloads, allowing developers to create smart data-driven applications, such as Datastream for BigQuery, a service that allows developers to easily convert data from transactional data to Repositories are replicated to BigQuery on the fly, and when paired with other similar tools, companies can act on data in real time without having to build infrastructure or handle related operations.

As these tools continue to improve, businesses will no longer need to separate transactional and analytical workloads until 2025.

8. Automate decisions related to cloud infrastructure

Google and Intel have worked together to design and build custom silicon, such as the Infrastructure Processing Unit (IPU), to deliver high performance and scalability for data-intensive applications.

Also, in the field of machine learning, Tensor Processing Units (TPUs) are currently used to power the world's largest, fastest, and most efficient supercomputers.

TPUs are up to 80% faster for large-scale training workloads and up to 50% less expensive than competing products.

These IPUs and TPUs, which are used to provide cloud services today, will allow Google to automate more than half of cloud infrastructure decisions in the next few years.

These processors support telemetry data and analytics using machine learning to proactively recommend the most suitable infrastructure based on the performance and reliability of individual workloads.

Once an enterprise specifies a workload, Google Cloud quickly recommends, configures, and delivers the best options based on the user's price, performance, and scale needs, outperforming any manually configured solution.

By the end of 2025, more than half of cloud infrastructure decisions will be made automatically based on enterprise usage patterns, and cloud thinking that considers hardware specifications will soon become a thing of the past.

9. Enterprises will be free to switch public cloud service providers

In recent years, enterprises will adopt multi-cloud strategies to spread risks and prepare for cloud migration in advance.

Google Cloud provides Anthos and related tools that allow enterprises to easily migrate between different cloud environments with consistent settings. For example, enterprises can use computing and data resources in other cloud environments through a multi-cloud management layer.

From databases to applications, businesses can run workloads on different public clouds, but manage them through a similar interface and execute them in a similar way.

For example, Google Cloud's Anthos can help enterprises manage Google Kubernetes Engine (GKE) clusters across dozens of Google Cloud regions, as well as manage other cloud environments, on-premises deployments, or clusters at the edge.

This means that enterprises can enjoy the benefits of Google Cloud tools, such as BigQuery for data analysis at any time, Spanner for distributed data, artificial intelligence and machine learning services for high-quality insights, and exclusive developer tools.

As more and more multi-cloud tools come online, by the end of 2025, more than half of organizations using public clouds will be able to freely switch between major cloud service providers.

10. Sustainable development will become an important development concept

In response to the global net-zero trend, Taiwan has passed the "Climate Change Response Act" to continue to promote the implementation of "2050 net-zero emissions", and will officially launch the carbon fee collection mechanism.

In addition to future developers focusing on how to create simple, fast and secure services in the most cost-effective way, sustainable development will also become a major consideration today.

It is expected that before 2025, 75% of developers will regard sustainable development as the main development principle.

Many IT executives today say they want to do more about sustainability, but don't know how to get started, and there's no way to measure the benefits.

But the situation is gradually changing. Google Cloud Carbon Footprint can help enterprises assess the relevant situation, make reports, and reduce carbon emissions of cloud services; enterprises can also access the Carbon Footprint information homepage directly from the Cloud console without additional settings.

In addition, the sustainable development team can obtain the required information through in-depth analysis and report the company's carbon emission information.

Businesses can now use Google tools to build applications and execute them in cloud regions with low environmental impact.

These tools are now free and open to use, allowing many developers to contribute to sustainable development in the enterprise.