-->

NEW EVENT: KM & AI Summit 2025, March 17 - 19 in beautiful Scottsdale, Arizona. Register Now! 

The Long- and Short-Term Impacts of AI Technologies

Article Featured Image

LLMs run on resource-intensive hardware GPU devices and data centers, requiring not just electricity, but also vast quantities of water to cool the systems. To address this insatiable need, Google and Microsoft plan to build nuclear power plants to fuel their resource needs. Depending on your viewpoint, that is either a sensible move by these tech giants or a profoundly frightening scenario.

Agentic AI is in its infancy and will take much longer to truly take hold and have the scale of impact that tech vendors and their investors would like us to think. But let’s not underestimate the long-term effects of agentic AI, as it’s likely to be far greater than we currently expect. The severe impact on many jobs and local economies requires thought and action to decide how to manage the fallout. Similarly, we must be aware that the current need to consume huge amounts of energy is not sustainable. We need to start planning and implementing more renewable energy resources.

Planning for the Future

Collectively, we need to consider when it’s OK to use agentic AI in the enterprise and when it’s not. Despite the billions of dollars driving the hype cycle and investors’ desire for profits, not everyone needs to jump on the bandwagon. Even if we are excited about using agentic AI in our organizations, technical, ethical, and business choices need consideration. Although LLMs are in vogue, they are overkill for most job automation activities because they are costly, complex, and poorly suited. In contrast, many cheaper, more efficient, and less energy-wasteful AI options are available.

It’s easy, mainly if you work in Silicon Valley, to believe everything can be explained through math and science. The limitations of today’s AI exist simply because we still need to build it bigger and more powerful. Along with many others in the tech sector, I disagree with that argument. AI technologies sometimes do a poor job—they will always be unsuitable for many human work activities, no matter how hard the sales pitch. In the real world, we humans face multiple daily situations that cannot be solved by applying predefined rules or procedures. AI is, or at least can be, very effective indeed, but by definition, it doesn’t possess, nor is it able to leverage, tacit knowledge. AI is not self-aware, nor does it have any form of intuition. There is a bumpy ride ahead when reality meets inflated promises. Even so, AI’s momentum shows no signs of slowing down.

The bottom line is that we need to pay attention to Amara’s Law and address concerns now, not once they have impacted society, not when it’s too late. It’s not that all AI is wrong or that people are leading the AI revolution with unsavory intentions (although there may be at least some truth to that). Instead, nobody has a detailed picture of what will happen other than knowing that agentic AI will have a considerable impact. Yet, as of today, few, if any, have any real plan to mitigate the potential worst outcomes, and those plans will come from concerned workers, employees, environmentalists, unions, and communities. If there is one thing I am sure about, they won’t come from tech companies and their investors.

KMWorld Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues