5 Ways AGI Could Unfold — And How to Prepare Now

Those who subscribe to the idea of AGI as a goal humanity is working toward — and may be close to achieving — base that belief on the following real-world dynamics:

• Progress: Creative outputs from GenAI models coupled with rapid improvements in multimodal models combining language and vision are driving fast improvements in AI.

• Investment: Investors have put billions of dollars into AI companies, even as smaller actors like DeepSeek achieve ambitious goals through more efficient methods.

• Hope: AGI capabilities are considered critical for solving seemingly intractable problems, like climate change and water scarcity.

The benefit of viewing AGI as imminent lies in optimism about a big, aspirational goal. Organizations that adopt this perspective should look out for innovations and improvements that exceed expectations. Closely follow composite and neurosymbolic AI, as these could mark the path to AGI.

Proponents of this perspective hold that real intelligence doesn’t exist in a vacuum, nor can it be simulated. Instead, it must be embodied in living beings and grounded in reality. Moreover, we shouldn’t underestimate the complexity of true intelligence, it most likely cannot be built.

The benefit of this belief is that it forces people to sharpen their understanding of intelligence and identify promising ways to overcome AI’s limitations, regardless of whether those innovations meet the definition of AGI. 

Those who adopt the “AI is impossible” view should validate their perspective by looking for AI hype to die down, advances to slow and attention to shift to new approaches. At the same time, entertain the possibility that you are wrong and reassess your beliefs every so often.

Those who see AGI as unpredictable argue that we do not know what it will look like because intelligence is a relative, subjective concept that is continually redefined and reshaped by context. As such, AGI could take one or several evolutionary paths that we cannot predict.

The appeal of this view lies in its openness to the possibility of alternative, non-human versions of intelligence. Yet its agnosticism also makes it difficult to know when or if AGI has arrived. By that logic, it may already be here and we’ll only know in retrospect. This makes it hard to plan for or to take action on, so keep actively following discussion of the topic.

Despite all the hype and investment surrounding AGI, some question whether achieving it is the right goal. It seems more important to have AI systems that can address difficult issues or solve challenging problems in a single domain than systems that promise cross-disciplinary capabilities.

Validation of this position will come if new AI issues beyond hallucinations and bias arise as a result of more cross-domain, continually evolving AGI-like systems. Similarly, an ever-lengthening timeline for achieving AGI could signal it’s the wrong path.

Some artificial general intelligence enthusiasts argue that the point is not to make AGI a reality, but rather to view it as a guiding ideal. Never reaching AGI could encourage investors and innovators to focus on developing solutions that address significant challenges and require generalist capabilities that exceed human capacity.

The benefit of this perspective is that it positions AGI as a progress accelerator that can solve problems beyond current human computational ability. Evidence that this position is taking hold will appear in the form of AGI’s inclusion in the mission statements of diverse companies and its integration into interdisciplinary projects.