Once, we used to compare computing power between Apollo 11’s guidance system and an iPhone. Updated, we now compare a USB-C charger with the rocket’s onboard computing; Apparently, it’s up to 563 times more powerful, and an iPhone could simultaneously land 120,000,000 Apollo 11s on the moon. Since the advent of Windows and Mac OS in the mid-1980s and two decades later mobile phones, the most striking element of this technological advance is the ease of use that has enabled almost anyone alive to understand, utilize, and benefit. NASA’s control room was reduced to a single voice command, ‘Take me home.’ Making something accessible to everyone is a literal definition of democratization. Now, another twenty years on, Artificial Intelligence and Machine learning, still mainly in the realm of specialists, are the targets of democratisers.
Technology has long been dealing with data collection, and so much obtained that the law of diminishing returns often takes effect. That is that analysis is overwhelmed by the quantity of data. But the critical issue is not really about the quality or amount of data. It’s about the ability of users to leverage that data for personal or corporate benefit. It’s been estimated that over half of all data is ‘dark data,’ not unusable but unused.
AI and machine learning are the alchemy that can turn data into golden insight. But, as in ancient times, alchemists, or specialized data scientists and technologists, are hard to find and expensive to keep. This means that AI development has till recently been the realm of larger companies and institutions as cost alone was the main bar to entry to using sophisticated data analytics development for SMEs. As we have seen, the essence of data democratization is to allow non-technical users ways of manipulating data to their own needs without assistance.
Airbnb is the textbook example of how AI helped a start-up grow so rapidly, allowing it to protect its reputation in a platform that almost invites fraudulent users. It uses over a hundred metrics to determine list rankings, trustworthiness, pricing, and other factors. Imagine the AI architecture needed to consistently make decisions that affect profitability and comply with a myriad of legal and regulatory requirements over seven million listings. The level of investment in machines and, more importantly, human expertise is equally awe-inspiring.
For Airbnb, the time and money invested in this technology was the essence of its business model. For smaller businesses, the advantages are self-evident. Mechanization of mundane, time-consuming tasks eliminates human error and the need for checking, targeted approaches, and automated conversations with clients/customers. Of course, all businesses, from small shops to aircraft manufacture, have their complexities. But the smaller the company, the fewer resources it has to invest in the technology to increase performance, productivity, and profit. Where cash flow is an issue, immediate ROI is essential.
By reducing costs and skill levels of entry, the democratization of data analytics in all its forms seems to be an apparent societal good. Google, Microsoft, and many others will help you set up AI for your needs and offer you processing. In addition, community groups provide advice and algorithms that may require only minimal tweaks to be used by another user. In short, the democratization of data analytics is underway and enabling non-specialist users to employ techniques previously barred to them.
Unfortunately, most societal goods are balanced by societal negatives. The newly installed AI technology might increase profitably massively within months. Still, if a ransomware attack then hits you, you have your IP stolen, or your billings hacked and diverted, it was all for nothing. The very openness of democracies is what leaves them vulnerable to others. Worse still could be the manipulation of the AI itself. Cybersecurity must be baked into the philosophy of democratization of AI, machine learning, and existing data analytic tools to continue their development safely.
But given that accessibility relies on sharing data to others out of your network and using sophisticated AI apps or other programs that may not have been tested virtually or developed by top-level practitioners, how do you achieve this? Surely democratization is the antithesis of the zero-trust paradigm? Segmentation, granular-level access controls, app/user authentication, and digital vaults seem to conflict. Yet, open democracies rely on police forces and armies to enforce rules and protect states. Zero-Trust is the component that fulfills that function in the world of democratized data.
Zero-trust works on micro-segmentation (the smaller, the better) and protection of every system and every step. It does this by combining user verification, exceptional logging, and other means to ensure accountability across data and systems access. In this way, users across an organization have access to the data and applications they need on a need-to-know basis without impairing their ability to deduce relevant insights resulting from blunter security tools.
Thus we should consider Zero-trust not as conflicting with the democratization of data analytics. Instead, it is the paradigm that will allow it to thrive.
HUB security is a holistic Confidential Compute solution made of military-grade HSM, key management, and cryptographic solutions for AI and machine learning applications, zero trust, critical infrastructure, finance, and blockchain security. HUB Security has developed a family of products that provide the highest level of enterprise security available on the market today. For example, the tamper-proof customizable box that sits in your location not only detects and prevents intrusion but ensures that data will not be intelligible if leaked by badly or maliciously written AI programs.