Are you ready to bring more awareness to your brand? Consider becoming a sponsor for The AI Impact Tour. Learn more about the opportunities here.
Everywhere you look, organizations are facing resource limits. If necessity is indeed the mother of invention, these resource constraints will give birth to new and disruptive innovations in 2023.
Every new wave of technology over the past 20 years has responded to the growth in data generated by people and things connected to a worldwide network. Organizations routinely use data from sources as diverse as client transactions, internal systems, external partners, the Web, social media and the physical world to inform their strategies, plans and operations. Just as every enterprise comes to embrace a data-driven approach to business, every employee in an enterprise will want to take greater advantage of all the data they can access.
To enable everyone to extract value from data, developers will need to build new interfaces to new systems that process and store data while working harder to preserve data privacy. Even though low-code and no-code interfaces have made data analysis easier, developers will enable new modalities of interaction in 2023 that are even more natural to human beings: vision and language.
Meanwhile, the explosion in data will drive the need for representations such as knowledge graphs that reveal patterns, relationships and connections among concepts and facts. At the same time, these intuitive interfaces to knowledge graphs will raise new concerns about privacy eroded by the exposure of hidden relationships.
The AI Impact Tour
Connect with the enterprise AI community at VentureBeat’s AI Impact Tour coming to a city near you!
New interfaces: From visualization to visual analytics
An image is the most bit-efficient way of making sense of data. Yet, the interfaces that drive many machine learning (ML) workbenches (such as Amazon SageMaker and Databricks) are notebooks that interleave text and code. Although you can embed visuals into notebooks, visualization is merely a representation of the computed output.
In 2023, visual analytics will drive new data science workbenches, making it possible to explore billions of rows of data, identify patterns and clusters and build hypotheses as easily as following directions or identifying points-of-interest on a map.
Transformative interfaces from smart speakers to smart assistants
Deep Learning models that enable phones and speakers to recognize and synthesize speech will combine with large language models (LLMs) that predict and generate complex text. Long after the notion’s debut in science fiction, everyone will speak with computers as easily as they do with friends and co-workers. But this journey will begin in the enterprise with data analysts. Just as transformative as visual interfaces, conversational interfaces will dwarf the usage of SQL and complex query languages for data analysis and data science.
Moreover, generative models trained on multi-modal data sets will enable developers to build agents that summarize, synthesize and generate text, speech, images and videos. In 2023, developers will adapt generative AI models to specific domains — from medical records or pharmaceutical reports to legal documents and scientific publications.
Today, generative models are unreliable for practical use — they do not correspond to the real world, do not have access to the Internet and do not claim to make true statements. In the fullness of time, we will use generative models along with “reality-checkers” that constrain their output. They will then augment our ability to analyze data and generate useful (and truthful) images, designs, charts, maps and other interfaces.
From databases to knowledge graphs
The knowledge graph’s time has come because data chains have become more valuable than data points. In 2023, enterprises will invest in representing data relationships as a network of nodes, links, weights and conditions. This will unlock intuitions and insights that are not evident in tables, charts or maps.
Some enterprises will lead by analyzing knowledge graphs constructed over domain-specific data sets. Foursquare, for example, is building a knowledge graph over its database of over 16 billion visits to over 130 million public points-of-interest around the world, extracting patterns and relationships in the movement of people across space and time to power location-based decisions.
From regulation-compliant to privacy-forward
To comply with new privacy regulations that require explicit consent for data collection and restrict data sharing, developers will turn to privacy-preserving methods such as clean rooms, differential privacy and homomorphic encryption.
In 2023, more enterprises will use clean rooms to query conjoined data sets without changing ownership of data. Using differential privacy (injecting noise into a data set and enforcing a privacy budget on queries), enterprises will protect the privacy of individuals while detecting patterns in aggregate.
Using homomorphic encryption, which preserves the form of mathematical operations over encrypted data, enterprises will build useful applications with encrypted data in memory. These techniques will become critical to many consumer applications.
I, for one, welcome the new regime of constraint-based innovation. Is this too much to ask from one year?
Vin Sharma is VP of engineering at Foursquare.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!