We often hear about AI reaching AGI, or Artificial General Intelligence, a form of AI that is not limited to one narrow task, but one that can learn, reason, plan, adapt, and solve problems across many different domains. If we look at the early expectations from AI pioneers, the idea is similar: that machines would eventually reach human-level intelligence, not only performing one narrow task, but learning, reasoning, planning, using language, adapting to new situations, and solving problems across many domains. Since then, AI enthusiasts have predicted that machines will reach AGI in the next 15 to 20 years, but we have not yet achieved this. However, we have made some significant progress if we look at the early days versus now. Some of the reasons this has not been possible are that it is more complicated to reach this level of intelligence than most people predicted, and we don't yet have enough computing power to support it. But one thing is for sure, this does not mean that it is impossible. If we look at history, we have truly progressed. Let's look at the current AI landscape. If you have been following the current AI trend, organizations are now deploying agents, and these agents are able to work alongside teams and learn from user interactions, making teams even more productive. Some examples include Shopify River tobi lutke
@tobi · May 9 Learning on the Shop floor Years ago I wrote about my apprenticeship in Germany. I dropped out of school at 16 and went to work at a Siemens subsidiary, where the most interesting people sat in the basement and used Delphi... Ramp, Atlassian Rovo, ServiceNow, and many others. I have also been working on something similar called Workpods, and it is amazing to see the progress we have made over the last two years since I started building agents. This is at the organizational level. If we look at personal life, almost everyone is using either GPT, Claude, or Gemini, which means two important feedback loops have already closed or are closing: personal life and work. The next frontier is robotics, and if robots are deployed and interact with humans in the real world, these models will continue to get better and better with personal data, organizational data and real world data. The question then becomes: are we going to reach an intelligence explosion phase where this models become soo good that they can be able to builds even better solutions, and the chain just continues?
If we look back in history, one of the reasons humans were able to dominate the world compared to other animals is that we can learn and accumulate knowledge across generations. That compounding effect is what separates us, and if machines are able to learn and accumulate knowledge in the same way, are we going to reach a point where they surpass humans in intelligence like human have compared to other animals and come to dominate the world? And if they do, will they simply become better tools for humans, or will they become the next dominant intelligence on Earth?