This blog post is the first in a series in which John Cho, Tria’s Chief Technology Officer, examines the impact of emerging technologies through his unique perspective.
Artificial intelligence (AI) is transforming many aspects of society, including the way people live, work, and create. But is this popular technology as limitless as it seems? With people sharing opinions on what AI is and what it can do, I’ve decided to flip the script and share my view on five things that AI is not.
AI is not without bias.
When we sequence and shape data for a large language model or even a large data set, we think we’re giving it good, accurate information. But the reality is that we inadvertently bake stereotypes into our statements and our training data. We embed a level of bias that we may not even be conscious of.
To counter bias in AI, tests should be developed before the code. We should also refine the precision in our prompt engineering techniques to enrich data sets with contextual explanations. Although we can’t eliminate bias completely, we can minimize its impact.
AI is not able to create on its own.
Yes, AI assists in the creative process. But it cannot create texts, images, and songs without our help. AI generates content based on what we tell it, and it infers and extrapolates based on that. In other words, it finds patterns in big datasets and uses that information to create content. Like Christopher Nolans’s 2010 sci-fi film “Inception” anticipated, today we’re planting ideas into models to create stuff.
AI is not a black box.
For many people, AI appears shrouded in mystery. To explain how it works in one sentence: AI involves the use of algorithms that analyze high-quality data to make statistically informed predictions and decisions. Those of us in the tech industry have the responsibility to demystify AI so that people understand the tool they’re using.
AI is not a panacea.
Wherever you look, the tech industry is touting AI’s ability to solve your problems. The expectations and hype are through the roof. However, AI is not a fit for every problem we’re trying to solve—at least not yet.
AI is not set-it-and-forget-it technology.
There is an industry term within the AI-solution space: “human-in-the-loop.” Because our models are not going to be 100% accurate, especially for complex solutions, people need to be a part of the adjudication process. On top of that, people need to be involved in the training of data, the labeling of data, and changing the testing parameters so the predictions can run correctly in a model–even during steady-state operations.
I’m as excited as the next CTO on the prospects of seeing how AI can assist us in areas like robotics, healthcare, and financial services. But make no mistake: People are here to stay. Yes, we’re moving from data-driven solutions to knowledge-driven solutions. But Albert Einstein said it best: “The true sign of intelligence is not knowledge but imagination.”