|
Getting your Trinity Audio player ready...
|
AI has made all kinds of things – from making images, to driving cars, writing music, and even doing well in medical tests. But how close are we to making a smart machine (AGI) that can think, learn, and use what it knows as well as, or better than, people?
It depends on who you ask and what you look at.
What Is Artificial General Intelligence (AGI)?
While the AI we have now, like ChatGPT or Google Gemini, is made for one job (like talking or knowing images), AGI is about a machine that can think like a person in every way.
AGI would not just be good at one task, it would be able to:
- Learn new skills without human programming
- Reason abstractly
- Understand context across domains (from math to philosophy)
- Improve itself autonomously
Behind the Numbers: Predictions & Possibilities
While the AI we have now, like ChatGPT or Google Gemini, is made for one job (like talking or knowing images), AGI is about a machine that can think like a person in every way.
- Surveys of AI researchers
- A 2022 expert survey published by Metaculus (a prediction platform) shows a median estimate of AGI by 2040–2050.
Source: https://www.metaculus.com/questions/5121/date-when-agi-is-publicly-known-and-acknowledged/
- The 2023 Artificial Intelligence Index Report by Stanford found that more than 50% of AI experts believe there’s a 50% chance of AGI by 2060.
Source: https://hai.stanford.edu/ai-index
- Investment growth
- The world’s AI market was worth $196 billion in 2023 and will likely jump by over 37% each year, hitting $1.8 trillion by 2030.
Source: https://www.statista.com/statistics/1365145/global-artificial-intelligence-market-size/
- A huge chunk of this is going into R&D for self-improving models, robotics, and brain-inspired architectures.
- Capability breakthroughs
- GPT-4 did great in hard tests like the Uniform Bar Exam and GRE. DeepMind’s AlphaFold cracked a tough biology problem.
- But these wins are still in narrow areas. AGI needs to be smart in a way that works for everything.
But these are still narrow achievements. AGI would require a unified intelligence that can generalize across all these domains and more.
So, When Will AI Be Smarter Than Humans?
Let’s separate this into two milestones:
1. When Will AI Match Human Intelligence?
Estimates vary:
- Optimists (like Ray Kurzweil) predict AGI will be achieved by 2029–2035.
- Skeptics believe we’re still decades away due to unsolved challenges in common sense reasoning, emotion, and physical embodiment.
Source: Kurzweil’s books: The Singularity Is Near (2005) & public talks/interviews.
2. When Will AI Surpass Human Intelligence (Superintelligence)?
This is where things get more speculative:
- Some theorists believe once AGI is reached, recursive self-improvement could make it rapidly smarter than us—within months or years.
- Others think it will take much longer due to ethical guardrails, compute bottlenecks, or lack of human-level embodiment.
What’s Holding AGI Back?
Despite recent leaps, we’re still facing major roadblocks:
- Contextual understanding: AI still lacks true “common sense.”
- Physical touch: Many AGI setups don’t have a form in the real world (like robots do).
- Ethics & match-up: Getting AGI aims to meet human values is a tough task (a field we call AI match-up).
The Bigger Picture: It’s Not Just About Smarts
The real question isn’t when AGI will be smarter than us—but How we’ll live with it.
Will AGI:
- Replace human labor?
- Create new jobs and industries?
- Pose existential risks?
- Accelerate scientific discovery?
AGI isn’t just a big tech leap—it’s a big step for us all.
Final Thoughts: Preparing for an AGI Future
Whether AGI comes in 10 years or 50, we’re starting to see what it can do. We, as people, leaders, and makers of rules, must get it, lead it, and get ready for it—not just in tech, but in right and wrong, in our social life, and in how we make money.
One thing is sure: the race for smarts has already started.
What do you think—will AGI beat us in our lives? Or are we running after a target that always moves?