A time to reflect and look forward

As we near the end of the year we take a look at key themes from 2022 and what to expect in 2023. Here are our top predictions for the year ahead.
Published 5 December 2022

As we rapidly head toward the end of the year, it’s naturally the time we begin to reflect on the year gone and begin to set our strategies and plans for the year to come. During 2022, we finally saw the back of lockdowns and restrictions imposed by the pandemic, and have gone on to grow our team, learned a lot through our customers, continued to develop our product, and successfully completed a new funding round. It’s been an exciting year!

We’ve seen a lot in the news about data privacy regulations and the advancements in AI showing huge potential in what technology is capable of. But, as consumers, we are beginning to ask questions about whether it’s right to be using certain data, whether we want our data used in this way, and raising concerns over ethics and bias. At Apheris, we feel organizations need to be better at making data practices more transparent and they should be accountable for how they collect, manage, and use data. But we also want to see the right data accessed and used for good to solve the challenges we see every day and to address the problems facing our planet.

We set out to help businesses solve the challenges associated with working on sensitive data and machine learning applications across organizational and geographical boundaries. As we look to the year ahead and how we solve these challenges we also took a minute to question what else might be on the horizon for us and our customers.

Here are our predictions for 2023:

  • Privacy regulation will continue to become more complex

    During 2022 we saw the UK state its intention to remove itself from the GDPR and write its own data protection regulation. California has CCPA and we expect other states to follow suit or perhaps we’ll finally see something at the federal level. Meanwhile, the EU Data Act is being revised and challenged and could be significantly different from what was originally planned.

    With increasing complexity across ever-changing regulations, businesses will struggle to move data across geographical boundaries, and will need to find alternative ways of working that avoid direct download of regulated datasets. The barriers for data sharing will become increasingly prohibitive and there will be substantial business risks for any data-driven global business that is operating on a central data infrastructure that is processing data from multiple countries. Organizations will have to find a way of navigating these barriers to remain at the forefront of industry.

  • AI regulation is coming

    With an increasing number of questions around how trustful, safe, biased, or ethical AI is, we expect to see AI become regulated. This will require transparency around how, when, and where the data used in developing the AI was captured. With more projects relying on cross-border AI, the creation of an AI Office to oversee the enforcement of the AI Act across geographies is being proposed and we may well see this happen.

  • Adoption of federated machine learning across geographical borders will be on the rise

    With increasing pressure not to share sensitive data and changing regulations across many geographies, centralizing data will no longer be an option for cross-border machine learning and AI initiatives. As a result, federated machine learning will be adopted at an increasing rate as organizations recognize the benefits of working with decentralized data as a way of overcoming many of the barriers they’re facing. However, initiatives will only be successful if appropriate governance is in place to protect commercially sensitive data and respect privacy.

  • Shift from model-centric AI to data-centric AI

    The adoption of MLOps principles and platforms has led to a shift from a focus on ML model architectures towards an emphasis on data pipelines, quality and automation when building and deploying ML applications. We expected this in 2022 and expect this will continue next year. With the questions mentioned in points one and two, emphasis must be on the data. Being able to unlock sensitive data that’s regulated, to make models more representative and more accurate, will be key for advancing AI in many fields from marketing and advertising through to healthcare and financial services.

  • Tougher market conditions will lead to consolidation

    With tough market conditions likely to continue throughout 2023, we will see consolidation in AI-first companies and investment will flow into AI and ML-based technologies where there’s proven value being generated. Finding ways to address strict regulatory requirements and protect IP, while removing bias and ensuring safety, will be a key driver for positive economic outcome.

  • There will be further commoditization in AI

    This will lead to standards and, for example, large language models, computer vision, and NLP applications will become easier to generalize and deploy. This commoditization will lead to a stronger foundation and standard set of building blocks for ML models that you can then customize for your market, region, or customer. It’s this customization that considers context and where businesses will be able to differentiate themselves.

  • Supply chain optimization will rapidly move up the manufacturing agenda

    With high energy costs set to stay for the foreseeable, energy-intense production processes will become non-profitable and there’s an increasing need for optimization within the, often complex, supply chain. This, combined with the new Supply Chain Act and ESG (Environmental, Social, Governance), companies need to find ways to track compliance with environmental standards across their full supply chain. Sharing this data or the digital twin of products and processes comes with heavy burdens in IT security, data governance, and IP protection, all of which will need to be addressed if manufacturers are to optimize supply chains and meet requirements imposed on them.

  • Businesses want to monetize their data products but fear losing control of their data

    Following on from 2022’s trend #4, a core principle of the Data Mesh Architecture is the establishment of Data Products. Teams and companies are looking for ways to monetize their data products but struggle with regulations, data governance requirements and the general fear of losing the control of their most valuable resource, data. To succeed and create additional revenue streams, there will need to be a mindset shift to one that looks for solutions to these challenges and works to solve them.

What are your expectations for 2023?

What are your 2023 predictions?

Do you have thoughts on data governance, addressing the needs of privacy regulation, cross-border AI, or monetizing data products? We'd love to discuss these with you.
Get in touch
Platform & Technology
Data & analytics
Share blog post to Linked InTwitter

Insights delivered to your inbox monthly

Related Posts