top of page

How Far Away Are We Actually from Human-Level AI?



MIT recently snagged an interview with Google DeepMind’s cofounder, Mustafa Suleyman, the prodigal son of Google DeepMind, who has since ventured off to cultivate his own digital Eden, Inflection AI. While one might have braced for a monologue of self-promotion, the conversation took an unexpected detour into the labyrinthine future of AI—specifically, the kind that doesn't just sit there but actually does things.


Suleyman sketched out the evolutionary roadmap of AI, which has had two major pit stops so far. You see, the first wave of AI was rather like teaching a parrot to recognize colors—a feat of classification if you will. Deep learning algorithms were the tutors, instructing our digital parrots to discern between a Monet and a meme, a symphony and a siren, a soliloquy and a shopping list.


The second wave has just arrived. Now our parrots have turned into poets, taking the colors they know and painting new rainbows in the sky. They're not just mimicking; they're creating. They're taking the data we feed them and cooking up entirely new dishes, as it were.


Now the third wave is what Suleyman is aiming for with Inflection AI. Imagine, if you will, a parrot that doesn't just squawk back at you but actually engages in meaningful discourse. A parrot that not only talks to you but also talks to other parrots and even people, coordinating, negotiating, and perhaps even gossiping. You'll simply whisper your heart's deepest desires into its feathery ear, and it will flap off to make it so, using whatever tools—or beak and talons—it has at its disposal.


You see, AI technology has long been a static entity, a genie that grants wishes but only when explicitly told to do so. But in the future, we’ll grant it a sort of free will, a discretionary power to act. It's a seismic shift, a genuine evolutionary leap. We're not just building tools anymore; we're nurturing entities with, dare I say, agency.


But how far are we really away from this future, and will it only be the result of an AI singularity, which essentially means that AI and technological growth become uncontrollable and irreversible with humans no longer having autonomy over technology?


Generative AI Just Cut the Road to Human-level Performance by 40 Years

In the twilight of 2022, ChatGPT emerged, not as a tentative step but as a full-blown sprint into the realm of artificial intelligence. It set a relentless pace for the AI space, with each month serving as a stage for new marvels in the theater of generative AI technology. March 2023 was particularly noteworthy, featuring no fewer than six pivotal advancements that you can read about in the diagram below.


Source: McKinsey & Company

If you thought that the third-wave parrot AI above wouldn’t happen during your lifetime, think again. Generative AI has shifted the game into high gear, making the cautious predictions of experts look rather quaint.


Experts at McKinsey have revised the timeline rather dramatically. The old estimate had us waiting until 2080 to have a heart-to-heart with an AI. Now, they're saying that by 2030—just a quick seven years away—AI could be as good at social interactions and emotional understanding as the top 25% of us humans. Just imagine an AI so adept in the art of conversation that if it had a body, it could blend right in at your weekend brunch. While McKinsey's focus is on those capabilities detailed in the chart below, that's just the starting point. However, it's not a stretch to say that a fully interactive, emotionally intelligent AI is not just a pipe dream; it's on the horizon.


Source: McKinsey & Company

Why Do Experts Think that Human-Level AI is Inevitable

In a universe where generative AI is already making waves, experts have peered into their crystal balls—or rather, their meticulously researched data sets—and declared a 50% chance that AI will outsmart us all by the year 2059. This is according to the 2022 ESPAI expert survey, and it begs the question: Why such optimism about interactive AI?

Let us ponder two facts:

  • Firstly, human intelligence is rather like a vintage car—charming but not upgradeable, unless we take a page from Elon Musk's playbook. His Neuralink startup aims to fuse our brains with computers, but let's be honest, they're still fumbling around in the neurological garage.

  • Secondly, machine intelligence is a different beast altogether. It's fueled by algorithms, processing power, and memory, all of which are growing exponentially, like a teenager who's discovered the gym. We've been rather adept at feeding these AI the algorithms they need to flex their computational muscles.

So, here's the crux: While we're stuck in cognitive first gear, machines are on the autobahn of intellectual development. Unless the universe throws a cosmic speed bump our way—a hard limit to machine intelligence, which we've yet to encounter—then it's only a matter of time before they zoom past us.





Comments


Want to Know When We Post?

How Far Away Are We Actually from Human-Level AI?

MIT recently snagged an interview with Google DeepMind’s cofounder, Mustafa Suleyman, the prodigal son of Google DeepMind, who has since ventured off to cultivate his own digital Eden, Inflection AI. While one might have braced for a monologue of self-promotion, the conversation took an unexpected detour into the labyrinthine future of AI—specifically, the kind that doesn't just sit there but actu ....

blurred text.png

Already Accessed Free Article !

bottom of page