Today, machine learning is at the core of how technology companies function. Tools like data visualization, smart workflows, and RPA (robotic process automation) enable organizations to prepare for the hyper-automation era, allowing different business applications to work together to drive efficiency.
To gauge how machine learning developments will shape the future of technology and secure critical data and digital resources from cyber threats, Toolbox spoke to experts to understand how the game-changing technology could transform the economy in 2021 and beyond.
Focus on Trust and Ethics
Jett OristaglioOpens a new window , Data Science and Product Lead of Trusted AI, at DataRobot
Artificial intelligence and machine learning are powerful technologies that have created endless possibilities, but in 2021, there will be a renewed focus across industries on AI Trust and Ethics. Thanks to infamous AI missteps in the private and public sectors, such as biased models parsing résumés or predicting educational outcomes, there is now massive public scrutiny around ethical AI, and rightfully so.Â
“Enterprises that take steps to embed ethical risk mitigation throughout the AI pipeline – from data prep through model development and production – can ensure that debiasing and other model evaluations take place on an ongoing basis, delivering true insights and responsible value to stakeholders.Â
Increasing the Use of Simulated Data to Train ML Algorithms
Nicolai BaldinOpens a new window , CEO and Founder of SynthesizedOpens a new window
A key trend shaping machine learning and the data ecosystem in 2021 is self-service, whereby businesses are demanding faster access to data and the ability to work collaboratively with trusted partners. This will help them produce products and services quickly to compete effectively.Â
“Simulated data looks, feels and behaves just like original data, but personal attributes have been removed about individuals so it is not subject to privacy regulations. Therefore, enabling businesses to collaborate freely without risk. Simulated data can also address bias, recognising when certain data points have been over- or under-represented within the original dataset and make corrections as necessary to avoid issues down the line.
See More: Building Trustworthy AI in 2021 and Beyond
Automating Tasks and Resolving the Complexity of Data Silos
Armand Ruiz GabernetOpens a new window , Director of Data Science and AI Elite Team at IBM Cloud and Data PlatformÂ
Three key ML trends going mainstream in 2021: talent, data, and trust. First, organizations will address their shortage of data scientists through automation like new AutoAI and one-button machine capabilities. Automating low-level tasks for data scientists will drive faster time to value.Â
“The second big trend will be resolving the complexity of data silos that exist in the hybrid cloud space. Finally, trust will be center stage as businesses look to deploy and scale AI. Organizations are realizing they must be able to trust their models and their business outcomes across the entire AI lifecycle. In 2021, they will continue to turn to tools like IBM Watson OpenScale and IBM Watson OpenPages that can help increase transparency, manage risk, and build greater trust in AI.
See More: Adaptive Insights CPO on Why Machine Learning Is Disrupting Data Analytics
Greater Role of ML in Banking Operations
Benoit GrangéOpens a new window , Chief Technology Evangelist at OneSpanÂ
The future of the banking sector is in the usage of more AI, machine learning, and biometrics and less passwords. Banks will combine machine learning with biometrics to provide new experiences, such as facial and fingerprint verification instead of passwords. One example we’re already seeing is banks leveraging machine learning to detect and read physical passports to allow for ID scanning. Customers use their smartphones to scan a government-issued ID and then take a selfie. The banks then leverage biometric facial comparison technologies with liveness detection to verify that ID is authentic and unaltered, confirming the individual’s identity.
Mark CrichtonOpens a new window , Senior Director of Security Product Management, OneSpanÂ
Currently a lot of banks and financial institutions have siloed data pools which can’t be pulled, however over the next year, it will be rare to see banks not using AI in an efficient way. When complex fraud detection models are able to be read and understood by people, then we firmly believe the power of AI will shine through across the banking industry.Â
Andy RenshawOpens a new window , VP, Payment Solutions and Strategy at FeedzaiÂ
AI and machine learning will become even more instrumental for banks who will have to meet their AML requirements as per regulations such as the EU’s Sixth Anti-Money Laundering Directive (AML6). By automating tasks that traditionally relied on manual work, such as know your customer (KYC) reviews, banks, and financial institutions will be able to improve accuracy and reduce their false positives rate.
But the future of AI and ML solutions is transparency and monitoring for potential bias – a problem that organisations can no longer afford to ignore. Goldman Sachs, for example, became the subject of an investigation in 2019 when consumers complained that its Apple Cards offered female applicants lower lines of credit compared to male customers. Those who adopt ML tools as part of their operations will need to take steps to ensure that the technology adheres to strong ethical standards.
Greater Focus on the Security of AI/ML Technologies
Nick McQuireOpens a new window , Senior Vice President, Enterprise Research CCS InsightÂ
While much of today’s focus is on applying AI technologies to cybersecurity, companies will soon put more effort in protecting their ML models and ensuring their algorithms are robust. By 2024, 50% of large organizations will deploy privacy-enhancing technology to support their ML applications. The value of customer data for insights and personalization is increasing, but so too are the privacy challenges, especially in the wake of greater regulation worldwide.Â
The tensions between these areas will continue over the next decade, especially in the development of big data and AI. Trust and respect for customer privacy are essential to long-term success, so companies have to invest in technologies such as data anonymization, differential privacy, synthetic data sets and homomorphic encryption for their machine learning projects.
See More: How To Bridge the Skills Gap Created Because of Automation
Role of ML in Automating Cybersecurity Tasks
Samantha HumphriesOpens a new window , Senior Security Strategist at ExabeamOpens a new window Â
Machine learning and automation have the potential to free up cybersecurity analysts to work on more critical/strategic activities by handling time-consuming tasks such as prioritizing security alerts, reducing false positives, and mapping devices to IPs. They can also enhance a security team’s ability to quickly detect attacker behavior that would otherwise require significant amounts of time to investigate manually. ML can also build out employee profiles, including their peer groups and personal email addresses, enabling analysts to identify insider threats far more quickly than was possible before.
Far from making junior positions redundant, ML will allow teams to hire more staff and help them to hit the ground running by putting powerful security tools and solutions at their disposal. In 2021 and beyond, the proliferation of these technologies will likely facilitate future evolutions within the sector, with new/different skill sets becoming more sought after by potential employers.Â
What’s your take on the best use cases of machine learning in the year ahead? Comment below or let us know on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!