Everywhere we look, products and services are being named and/or described using tech-related acronyms and buzzwords—to the point that they are losing their meaning and value. We see words like AI (artificial intelligence), ML (machine learning), blockchain, bots, analytics, data warehouse, and others in the names of products without any explanation as to what they mean and why we should care. While these technologies are important, potentially transformative, and valuable—some even having new types of economies based on them—we run the risk of making them seem trite when they are attached to products or services without any understanding of how they make the offering better than the competition. The key for us is not to ignore or disparage these technologies, but to look at the products and services they power to understand how (and also if) they improve them.
AI – Artificial Intelligence
A common definition of AI, from Wikipedia, is that the term describes: machines that mimic “cognitive” functions that humans associate with the human mind, such as “learning” and “problem solving”[i]. This is an extremely broad and ambiguously defined discipline, including a number of those technologies listed above—often in combination with each other. AI is commonly used in software to do repetitive, logic-based work that is comprised of well-defined steps or processes. A side note is that some of these functions can become so commonplace over time that we don’t even consider them AI anymore, like optical character recognition.
In the world of background screening, there are some ways AI helps today and will help more in the future. Bots have been employed for years to mimic a human logging into a site to pull information in place of a person. This has decreased turnaround times and can provide a more consistent result format. We also see machines performing algorithms on screening results to determine whether other locations should be searched or whether there is adverse information to report. These algorithms are based on the processes that a person would perform and can be done faster, with less bias (assuming bias isn’t programmed into the algorithm—a problem for another blog post), and more consistent results.
ML – Machine Learning
ML is a specialized, advanced form of AI. In this case, instead of the computer being fed a fixed, logic-based algorithm that it must follow, the machine continues to develop, change, and improve the algorithms based on data and experience. This is a particularly difficult area to get right, which is why most of the examples we see so far are very limited in terms of scope. ML is also prone to bias even more so than AI because the algorithms are learning and improving based on the data and outcomes. When the outcomes are arrived at based on some form of individual bias, the machine will then be learning and improving to incorporate this bias.
Some common examples of ML in our world today include recommendation engines like those used by Amazon and Netflix, speech and image recognition tools, and self-driving cars like Tesla and Waymo. This is a largely new area for screening software, as most of the algorithms in use are based on existing processes and matrices. However, as the amount of data increases—specifically related to the outcome of a search—it will become more practical to train the machines to build their own algorithms, which will not only improve accuracy and efficiency, but will hopefully help us to recognize new and valuable relationships in the data that help us to better predict outcomes.
In my opinion, Blockchain holds the distinction of being the simplest concept with the most straightforward definition that has the most complicated, headache-inducing implementation of any technology. According to Wikipedia, “A blockchain is a growing list of records, called blocks, that are linked together using cryptography.”[ii] [A fuller definition from the same article, “A blockchain is a decentralized, distributed, and oftentimes public, digital ledger consisting of records called blocks that is used to record transactions across many computers so that any involved block cannot be altered retroactively, without the alteration of all subsequent blocks.”] There’s a lot more to it, including the timestamping of each block, the chaining together of each immutable block, and the management of a distributed ledger through peer-to-peer networks. In the end, it’s highly secure, decentralized, fully audited, and open.
The most publicized use of blockchain as a technology is in cryptocurrencies like Bitcoin and Ethereum, but that’s only one way to use it. Some other examples are: secure data sharing (e.g., medical data or background checks), supply chain and logistics (e.g., integrations with data providers), and personal identity security (i.e., keeping your identity and credentials secure to share in an audited way).
The bottom line is that these buzzwords do not make the products or services instantly or intrinsically better (what the marketers want you to think), nor do they make them suspect or lessen their credibility (what their competition want you to think). The key questions when evaluating an offering or trying to determine whether your technology should employ one of these technologies are:
- How is this product better because of this technology? (Faster, more efficient, more scalable, more accurate, etc.)
- Does it solve a real problem? (Insecure data sharing; algorithms that are constantly in need of updating; people being asked to perform repetitive manual tasks; etc.)
If you can answer those questions—or the provider can—then you’ll be able to make valid comparisons between offerings and/or development approaches that will help you to employ the right technology for your business, and not just the coolest sounding one. This is for the record.
For The Public Record is a monthly blog featuring thought leadership from the most seasoned experts at ClearStar, across all functions of the background screening process. Click here to subscribe.