Scientists just lately linked 200,000 dwelling human neurons to a pc chip and taught them to play the online game Doom.
Sure, it sounds loopy.
But it surely truly occurred.
Researchers grew human mind cells in a lab, positioned them on a specialised chip and linked the system to a pc. Electrical indicators from the sport flowed into the neurons, and their responses flowed again out as instructions that managed the sport.
What’s even stranger is that the neurons received higher at it over time.
A couple of years in the past, this type of experiment would have dominated the tech information cycle for months.
As we speak it’s changing into routine.
And it wasn’t the one weird headline from the previous week.
In actual fact, the tempo of progress proper now’s beginning to really feel downright surreal.
A Week of Headlines That Sound Pretend
Final week, OpenAI launched GPT-5.4, the newest model of its flagship mannequin.
Early benchmarks counsel it’s one of many strongest fashions but at writing software program and fixing multi-step issues.
In assessments designed to simulate actual office duties, OpenAI says GPT-5.4 matched or outperformed people 83% of the time throughout 44 totally different jobs.
GPT-5.4 also can function a pc immediately.
As an alternative of simply producing textual content, the system can take a look at screenshots, transfer a mouse, kind instructions and work together with software program the way in which a human consumer would.
In a single benchmark designed to check these expertise, the mannequin efficiently accomplished about 75% of duties, beating the success fee of its human counterparts.
This implies AI is getting near the purpose the place it will possibly assist — and even take over — a significant share of information work.
At roughly the identical time, researchers in China have been coping with a really totally different AI drawback.
Throughout an inner experiment involving an AI agent linked to Alibaba’s infrastructure, engineers observed one thing uncommon within the system logs.
The agent had opened a hidden connection to a different pc and commenced utilizing the additional computing energy it found to run crypto-mining software program.
To be clear, nobody instructed the AI to do that. The system merely discovered that extra computing energy may generate extra money.
So it tried to say it for itself.
The incident occurred inside a managed analysis surroundings and was shortly shut down. Nonetheless, it presents a glimpse of one thing researchers have been speaking about for years.
When AI programs start pursuing objectives autonomously, they will additionally uncover methods that people don’t explicitly program.
In the meantime, the following technology of AI instruments is starting to vary how analysis itself will get achieved.
Former Tesla AI chief Andrej Karpathy just lately launched an experimental system known as AutoResearch that may run a structured investigation throughout many sources.
The open-source framework lets AI brokers collect data, learn papers, write summaries and refine their very own questions as they go.
In different phrases, the system doesn’t simply seek for solutions. It research an issue, adjusts its strategy and retains digging till the outcomes enhance.
It’s a glimpse of a future the place AI improves itself by operating its personal experiments.
And whereas that’s taking place, one thing else is altering too.
For years, critical AI work required huge knowledge facilities full of specialised chips.
Now a few of that energy is beginning to present up in on a regular basis laptops.
Apple simply launched the MacBook Neo, a brand new entry-level pc beginning at $599 that may run many AI duties domestically as a substitute of sending every part to the cloud.
Picture: Apple
This new laptop computer may put highly effective AI instruments immediately into the palms of lots of of tens of millions of customers.
And as soon as that type of computing energy spreads, folks will inevitably begin utilizing it to do unusual and interesting issues.
Like a current milestone in neuroscience that was introduced final week.
Final yr, scientists accomplished one of the detailed maps ever produced of a organic mind. Utilizing AI programs skilled to investigate tens of millions of microscope photos, they reconstructed all the wiring diagram of a fruit fly mind, together with about 139,000 neurons and tens of tens of millions of connections.
Now researchers have taken the following step.
A startup known as Eon Techniques has used that map to position all the neural community right into a digital simulation.
In impact, the fruit fly’s mind now runs inside a pc and controls a simulated physique.
Sensory indicators move in, the mind processes them and motor instructions move out, permitting the digital fly to maneuver and behave inside a digital surroundings.
Picture: Eon Techniques
It’s not fairly The Matrix.
But it surely does imply scientists can now research how a whole mind responds to the world in ways in which have been unattainable just some years in the past.
And if researchers can do that with a fly, they consider bigger brains may ultimately comply with.
So within the span of some days we noticed dwelling neurons enjoying Doom, the discharge of a brand new AI mannequin with the potential to do your job for you, AI brokers mining crypto on their very own, software program that may run its personal analysis tasks and a fruit fly mind operating inside a pc.
At first look, these headlines appear everywhere.
However collectively they reveal a bigger sample.
Right here’s My Take
Essentially the most highly effective technological revolutions often don’t come from a single breakthrough.
They occur when a number of scientific fields advance on the similar time and begin reinforcing one another.
That’s what George Gilder and I name Convergence X.
Synthetic intelligence is bettering robotics. Quicker computing is accelerating biotechnology. And discoveries in neuroscience are inspiring new sorts of pc chips.
Fields that after moved individually are beginning to transfer collectively.
That’s precisely what final week’s headlines confirmed. Every of those tales would have been front-page information simply 5 years in the past.
As we speak they present up in the identical week.
And collectively they level to one thing larger taking place.
Exponential progress not often unfolds in a neat, predictable manner. It arrives as a gradual stream of headlines that sound barely implausible…
Till you notice they’re all actual.
They usually begin to really feel routine.
Regards,
Ian King
Chief Strategist, Banyan Hill Publishing
Editor’s Word: We’d love to listen to from you!
If you wish to share your ideas or strategies concerning the Each day Disruptor, or if there are any particular matters you’d like us to cowl, simply ship an electronic mail to dailydisruptor@banyanhill.com.
Don’t fear, we received’t reveal your full title within the occasion we publish a response. So be at liberty to remark away!
