Managing AI expectations
After having had some time to reflect on everything I learned from the nexxworks AI tour in Munich, I want to share my top three takeaways from all the fantastic company visits:
1. We’re not there yet: The potential of AI is incredibly exciting, but there’s still a long way to go before it becomes fully reliable and seamlessly operational. We face challenges like AI hallucinations—where it generates false or misleading information—along with concerns about transparency, data privacy, compliance, and copyright issues. Another complex hurdle is developing true reasoning capabilities in chatbots like ChatGPT or Gemini. While companies like OpenAI claim significant advancements, AI consistently reasoning like a human remains a formidable challenge.
2. The energy problem: One of the biggest obstacles for AI is its massive energy and water usage, along with the CO2 emissions generated by data centers. For example, Microsoft is currently spending $1 billion a week on its data centers. It’s not surprising that companies like Microsoft, Google, and Amazon are investing in nuclear energy, as they did recently. Dublin even denied Google’s request to build a third data center due to the energy strain it could cause. So, it’s not just about maturing the AI software—its hardware must evolve to reduce its environmental and energy impact as well.
3. Disappointments shouldn’t stop experimentation: Amara's Law, coined by futurist Roy Amara, states that people tend to overestimate the short-term impact of new technologies while underestimating their long-term effects. That’s exactly what’s happening with AI today: expectations are too high in many areas, and people will likely be disappointed in the short term. But once we reach a tipping point, AI will start doing much more than anticipated. Companies that don’t prepare or experiment now—because they’re discouraged by AI’s current limitations—will face serious trouble when we hit that point.
2. The energy problem: One of the biggest obstacles for AI is its massive energy and water usage, along with the CO2 emissions generated by data centers. For example, Microsoft is currently spending $1 billion a week on its data centers. It’s not surprising that companies like Microsoft, Google, and Amazon are investing in nuclear energy, as they did recently. Dublin even denied Google’s request to build a third data center due to the energy strain it could cause. So, it’s not just about maturing the AI software—its hardware must evolve to reduce its environmental and energy impact as well.
3. Disappointments shouldn’t stop experimentation: Amara's Law, coined by futurist Roy Amara, states that people tend to overestimate the short-term impact of new technologies while underestimating their long-term effects. That’s exactly what’s happening with AI today: expectations are too high in many areas, and people will likely be disappointed in the short term. But once we reach a tipping point, AI will start doing much more than anticipated. Companies that don’t prepare or experiment now—because they’re discouraged by AI’s current limitations—will face serious trouble when we hit that point.
