5 Expert Predictions for AI/ML and Data Science in 2022

In 2021, the trend toward increasing innovation, adoption, and investment in AI and data science applications continued unabated, even as industry growing pains around issues like ethics and bias persist. For many, 2021 was an exciting year when data science possibilities matured into more clear-eyed views of the potential positive business impacts. There’s never been a more exciting—and important—time to be a member of the data science community.

Where does this trajectory lead in 2022 and beyond? To find out, we asked experts at Anaconda and other leading industry voices for their predictions on the year ahead. Here are their views on what to watch for in the coming year—if you want to hear more of our predictions for next year, tune in to our webinar on Dec. 15, where leaders from Netflix, Meta, and Wikimedia will join our very own Peter Wang to discuss what’s next for AI and ML in the new year.

Delivering on the promise of ethical AI

Ethics challenges in AI have never been more in the public consciousness, yet there is still much more progress needed in this space. For the most part, AI workstreams haven’t fundamentally changed. Instead, some companies are building tools to help detect ML model drift, while people in the industry are being asked to police themselves with these and other tools. Kevin Goldsmith, Anaconda Chief Technology Officer, believes that leaving the ethical implementation of AI up to individuals and their organizations is ineffective and that, moving forward, we’ll see more regulation and industry-wide governance around the use of AI and personal data. This will likely happen through government interventions and new wide-scale standards vital to leveling up the ethical qualities of AI across the board. While fully adopting these standards will take time, self-monitoring won’t cut it in 2022 and beyond.

At the same time, this doesn’t mean that practitioners should renounce all responsibility for practicing ethical AI. James A. Bednar, Director of Technical Consulting at Anaconda, believes we can’t ignore that the practice of ethics starts with people. As the field continues to grow and some capable researchers and practitioners are making strides to create a more fair and effective AI, very few people truly understand all of the aspects required to use these tools responsibly and effectively. The new year should see renewed efforts to equip individuals with the resources they need to understand the application domain and ML algorithms, especially as cloud computing and AutoML technologies continue to become more accessible.

The end of innovating for the sake of innovation

Enterprises naturally focus on the opportunities that the newest technologies and tools can bring to their businesses. As a result, considering how different technologies will actually affect users has often been a secondary consideration. This will change in 2022 and beyond, according to Lucia Gallardo, Founder and CEO of the socio-technology development lab Emerge. As a broad set of stakeholders advocate for greater oversight of how technologies impact individuals and society, this will require us to step away from traditional thinking about sustainability, inclusion, and impact, and pivot these considerations from intended outcomes to long-term embedded strategies. Ethical concerns will accelerate investments guided by environmental, social, and corporate governance (ESG) criteria and encourage efforts to improve how we measure impact.

Addressing training data and security challenges with transparent traceability

Tasked with meeting the demand for sophisticated data-centric products, developers often turn to training data “found” on the internet. For years, the ethical and legal implications of this common practice have gone mostly unaddressed, but 2022 may mark a turning point, according to Stan Seibert, Sr. Director of Community Innovation at Anaconda. Questions about rights to use different datasets and the licenses that might govern them have become even more complex with the introduction of AI-powered coding assistants. These products use public source code itself as the training data. While most public source code has a license attached, it’s not clear how these licenses apply to training data for a model. Our legal understanding of licenses and copyright will need to evolve to account for these new machine learning use cases. Answering questions like these, and establishing guiding precedents, will be a focus for 2022, according to Stan.

Aside from licensing issues, the mix-and-match nature of modern software development can also pose security challenges. With small snippets of code often embedded in larger projects, it can be difficult to track the provenance of the code, especially in the open-source software world. As issues around traceability garner more and more attention, with the federal government weighing in on the value of a Software Bill of Materials (SBOM) in an executive order earlier this year, customers will demand to know where all the code running in their systems came from. This new standard will bring changes to the ways that practitioners develop code, according to Stephen Nolan, Anaconda’s Sr. Vice President of Product. In 2022, developers will begin monitoring application and code security more carefully, as well as implementing security measures across all stages of code development. It’s unreasonable to expect developers to be constantly monitoring for vulnerabilities, which is why solutions like conda signature verification will be useful for practitioners and enterprises alike in the new year.

Further broadening the audience and use cases for Python

As enterprise data science teams multiply, more and more are using Python. The language recently overtook Java and C for the No. 1 slot in the TIOBE Index, a measure of the popularity of programming languages, and the demand for software developers is projected to grow 22% from 2020 to 2030. Saundra Monroe, Director of Product Management at Anaconda, recognizes that both novice and experienced developers are drawn to Python due to its dominance and simplicity, and thinks that one area of focus for the language as a whole in the new year will be improving the paths available for beginners to become capable coders.

According to our experts, besides expanding to more coders (whether professionals, students, hobbyists, or somewhere in between), Python will also continue expanding to newer use cases beyond data science in 2022. Stan believes that, for use cases like microcontrollers and IoT devices, where other programming languages have typically dominated, we’ll see growth in the adoption of Python due to the rise of MicroPython and CircuitPython. Taking the question in a different direction, Joseph J. Currenti, Sr. Technical Account Manager at Anaconda, and Lucia said they expect Python to be used more in game development as developers look to AI to create more immersive gaming experiences.

Empowering the community through standardization and education

Recent years have seen a growing push for standardization and cohesion from different tools and communities in the software development space. Sebastián Ramírez Montaño, Staff Software Engineer at the AI-powered software platform Forethought, believes this momentum will only increase in 2022 as developers seek to build packages supported by many tools, such as editors and cloud providers. Examples of this principle in action include Python-type annotations, and async/await, among others. Standardized tools are more accessible for developers to use, ultimately helping grow the number of communities working and developing with Python.

Another key to growing the Python community in 2022, according to Sebastián, will be improving education around elements like async and await, concurrency, threads, and task-local state, because using them correctly is key to getting better throughput performance. These concurrency-related features will be critical as programmers increasingly use distributed systems, microservices, and integrations with other tools to accomplish tasks like connecting ML serving systems with Python APIs.

Looking ahead to an exciting new year

From working to mitigating bias in AI models to the dominating rise of Python, it’s been a year of growth for the data community. We’re excited to see further progress in the field and more widespread adoption among thousands of different use cases in the new year in the enterprise setting. There truly is something for everyone in data science, and we’re proud to be a part of that! Join our webinar on Dec. 15 at 2 PM EST/11 AM PST to hear more predictions from industry leaders.