Top Technology Trends 2022: What to Expect in the World of Software Development (Part 2)
In our previous blog post, we talked about web and mobile development and what trends will dominate them in 2022. Today we’ll continue discussing the biggest IT trends and ways companies can use them to their benefit.
According to Expert Market Research, the global market for database management systems will reach approximately $125,6 milliard by 2026. As for the most popular database management systems, the top three are Oracle, MySQL, and Microsoft SQL Server, with PostgreSQL and MongoDB following them.
There are also several interesting database trends to observe this year:
- Cloud-based DBMS: experts predict that more data storage system vendors will switch to cloud-like on-demand elastic pricing in order to efficiently compete with cloud storage providers. The on-demand elastic pricing proved to be a win-win for both customers and vendors and will most probably turn into a big trend.
- AI & ML: we can definitely expect to see more of AI and ML in data management activities, such as data identification and categorization. No need to remind you that the use of Artificial Intelligence can greatly boost the accuracy and speed of processes and database management is no exception.
- Open-source databases: Gartner predicts that this year, more than 70% of new in-house applications will be created with the use of open-source database management systems or cloud-based database management platforms as a service.
Artificial Intelligence and Machine Learning
In 2021, SoftTeco opened a Machine Learning and Data Science department so we are naturally interested in the latest trends within the field. Before looking at several biggest ones, let’s look at some general facts first.
Python remains the number one programming language for Machine Learning with 57% of data scientists and ML developers using it. It is closely followed by C++, R, Java, and Julia. As for the most popular ML libraries and frameworks, the top three are Scikit Learn, PyTorch, and TensorFlow.
It’s also worth looking at the global AI revenue forecast ($ millions) as per Statista:
- Static image recognition, classification, and tagging: 8,097;
- Algorithmic trading strategy performance management: 7,540;
- Efficient and scalable processing of patient data: 7,366;
- Predictive maintenance: 4,680;
- Object identification, detection, classification, tracking: 4,201.
In general, Artificial Intelligence is now an incredibly hot topic. More and more companies are rapidly adopting AI to improve their services: a 2021 report by McKinsey states that 57% of companies are using AI in at least one business function. In 2020, the number of companies using AI was 45% only. Gartner predicts that by 2025, 70% of companies will have operationalized AI architectures and in general, spendings on AI implementation are expected to grow by approximately $204 billion in 2025.
Language modeling with GPT-3 and GPT-4
In 2020, OpenAI (a San Francisco-based AI research laboratory) released GPT-3 which stands for Generative Pre-trained Transformer 3. This autoregressive language model deploys deep learning to produce human-like text and is comprised of 175 billion parameters (GPT-2, in comparison, comprises only 1.5 billion). The most popular use cases for such models are email generation, website building, chatboxes, and basically everything that involves creating copy. Should writers be worried? Probably.
Now, it’s interesting how quickly OpenAI has been releasing new versions of GPT: the first one was rolled out in 2018, GPT-2 was launched in 2019, and GPT-3, as already stated, in 2020. So it’s kind of safe to assume that GPT-4 will be coming into play in early 2023.
During his Q&A session at an AC10 online meetup, Sam Altman (CEO of OpenAI) spoke about GPT-4 and what we might expect from it. The main takeaways are that the new GPT version will not be much bigger than GPT-3 by the number of parameters. However, GPT-4 is planned to use more compute resources and will be focusing more on coding. One more interesting thing that OpenAI works on is how to train smaller AI models to deliver accurate results and resolve complex problems. While it’s a well-known fact that a big AI model equals higher accuracy, some research groups prove the opposite. So training of smaller models has become a point of interest for OpenAI researchers and something to work on in the future.
Natural language processing (NLP)
Last year, data was at the center of attention and companies placed a special focus on unstructured data. Unstructured data is data that has not yet been processed in order to be stored in a relational database and examples of such data are emails, social media posts, or rich media.
Considering the fact that IDC expects the volume of unstructured data to form about 80% of all the data by 2025, it’s natural that companies are looking for efficient ways to structure and understand this data in order to gain valuable insights from it. This is where natural language processing comes into play.
Researchers at Expert.ai predict that NLP will become one of the hottest AI-related trends in 2022 and that this year will be the year of natural language-enabled enterprises. That means enterprises will either consider or adopt an NL-based strategy in order to make the most out of available unstructured data.
Low-code and no-code AI
In the last few years, low-code and no-code development have become incredibly popular. These development methods imply using user-friendly platforms with a drag-and-drop interface to create software products and yes, these tools can help you build an AI model.
Now, the demand for AI is insanely high today but there are simply not enough specialists: according to the report by Forbes, 83% of companies consider AI to be their strategic priority but state there are not enough data science talents. As well, hiring an AI professional may be too expensive for a company and thus, a low-code AI platform may become a good alternative. In this way, a business gets a chance to optimize its processes with Artificial Intelligence while saving costs and adopting an AI solution in a quick manner.
It’s interesting though that no-code AI is still less popular than people’s interest in learning Machine Learning and AI in general, according to Google Trends. So it’s safe to assume that no-code AI will not replace data scientists and AI professionals for at least a good while.
Cybersecurity has always remained a hot issue and now, in the era of the pandemic, it became especially acute. For instance, a report by Cybersecurity Ventures states that global cybercrime costs will reach approximately $ 10.5 trillion by 2025 (compared to $3 trillion in 2015). As for the cost of cyber attacks, the average breach cost due to compromised credentials was $ 4.37 million, according to the report by IBM.
With these numbers in mind, let’s have a look at what’s trending in the world of cybersecurity and what to pay attention to if you want to keep your software safe.
The zero-trust approach is one of the biggest cybersecurity trends for 2022 and there is a good reason for that. Companies that applied a zero-trust approach experienced $ 1.76 million less of the average cost of a breach, and we can all agree that the number is quite impressive.
Unlike a traditional approach towards user authentication and verification, the zero-trust approach implies trusting nobody. With that said, here are the main principles of this approach:
- Least privilege: users have only as much access as they need and this access is granted through managed user permissions.
- Continuous validation: since attackers can be both outside and inside the network, no one can be trusted. Hence, all users and all machines are to be regularly monitored and re-validated.
- Control of device access: there is a certain number of devices that can access the network and all devices are carefully monitored and checked.
- Microsegmentation: security perimeters are segmented into small zones so that separate parts of the network require separate access.
Yes, a zero-trust approach towards security sounds rather mundane - but it’s better to invest in it once than deal with massive financial losses in the future in case of a breach happening.
5 biggest cyber threats
Another important thing for organizations to remember is what cyber threats are expected to be the biggest ones in 2022. We’ve written an article on the issue so please see it for a detailed explanation. For now, here is a quick overview:
- Ransomware: an attacker encrypts victim’s files and then demands a fee to give access to the files back;
- DDoS: an attacker overfloods one’s website with traffic and makes it stop working;
- Phishing: an attacker sends malicious emails with an aim to make a user click the link and disclose sensitive information;
- MITM: an attacker steals sensitive information by inserting himself “between” a sender and a receiver of this information.
- Trojans: an attacker disguises malware as legitimate software to trick a user.
Of course, these are not all possible threats and vulnerabilities to watch out for - we highly recommend checking the official OWASP Top 10 documentation to learn about the biggest risk areas and how to mitigate them.
Evolution of threats
As if the existing threats were not enough, the bad news is - they keep evolving. So in 2022 and on, companies need to review their cybersecurity strategies and adapt them to the changing nature of cyber threats.
Examples of such evolution are triple extortion ransomware, spear phishing and whaling, and new DDoS attack types. As technology advances, so do cyberattacks and hence, companies need to remain alert and take preventive measures to detect risks at the earliest stage.
Employee security training
One of the best ways to prevent and mitigate cybersecurity risks is employee security training and in the last few years, companies began to invest heavily in it. A report by Cybersecurity Ventures states that the global spending on security awareness training will reach $10 billion in 2027. A fun fact: a few years ago, security training was the most underspent budget item regarding cybersecurity (and look where it is now!). So if you have not yet considered implementing security training in your organization, now it’s probably time to do so.
As you can see, all areas of software development undergo certain changes, which are mostly aimed at the convenience of development, security, and great user experience. We hope you enjoyed our report and we highly recommend adopting certain trends (especially related to cybersecurity) as their early adoption may result in highly beneficial results in the long run.
SoftTecoView all articles by this author.
Great content! Keep up the good work.
hi, This blog is informative and helpful. it really amazing.