@aidecipher: Demis Hassabis did a back-of-envelope calculation in a meeting. ⠀ Started asking simple questions. How many proteins exist in nature? 200 million. ⠀ How many computers would we need to fold them all? ⠀ How long per protein? If we folded one every 10 seconds? ⠀ Somewhere in the middle of that calculation, fiddling on his phone, it hit him. ⠀ It would be possible in a year. ⠀ That's when the strategic insight arrived. Why build servers? Why build databases? ⠀ Why build all the infrastructure of a traditional service platform? ⠀ If we could fold everything anyone could ever request or want, we could just put it all on a database for free. ⠀ Free for all scientists in the world to use. ⠀ That moment changed biology. ⠀ AlphaFold didn't become a service platform where researchers request protein structures. ⠀ It became a free resource. Almost every known protein structure predicted and available. ⠀ No paywalls. No gatekeeping. No infrastructure bottleneck. ⠀ Just open science. ⠀ Here's where this differs from how technology usually gets deployed. The default path is: build the service, monetize usage, constrain access. ⠀ Hassabis chose: solve the entire problem, release it free, accelerate science. What changes when you shift from "build a service people pay for" to "solve the problem completely and release it free"? ⠀ Credit: Demis Hassabis | Cleo Abram ⠀ 👉Follow or You May Never See Us Again ⠀ #AlphaFold #Biology #Science #Open #Breakthrough

AI For People Tired of Noise

Open In TikTok:

Region: US

Thursday 04 June 2026 00:55:03 GMT

21917

503

Music

Download

No Watermark .mp4 (2.32MB) No Watermark(HD) .mp4 (2.3MB) Watermark .mp4 (2.15MB) Music .mp3

Comments

EMBER & YUKI :

Dude, is he claiming that he solved the protein structure of disordered, membrane proteins, and protein assembly— most scientists know that protein structures are not stationary, they are “molecular machines” (lookup kinesin walking the microtubles, if you are interested.) While alpha-fold is pretty good, we are long way from solving all protein folding problem and no where near what he talking about. Do we really need to inflate when the science is actually pretty good where is it right now? I had a colleague who fell for this PR and wasted a lot of resources folding sequences (multi-mers, for example) that just can’t be done. Hogwash.

2026-06-04 01:26:08

Fernando M :

without ai this would have taken thousands of years

2026-06-05 07:54:12

Carter :

And the first caveman to pick up a stick assumed he had mastered weaponry

2026-06-05 10:37:53

brian :

I ran openfold on PlayStation 3 when it first came out

2026-06-05 04:31:43

Dad Top Tips !!! :

Now tell it to apply that to predict what other intelligent species could possibly look like

2026-06-18 22:20:17

Chimeras :

You can "predict" what you want, it doesn't mean it's active though.

2026-06-05 01:33:50

Homosapienish :

Yes

2026-06-05 18:08:38

pervywizard6 :

Instead of writing a novel, let's build the tower of Babel. oh thanks I needed that.

2026-06-05 18:13:30

authorAndrew :

in that meeting someone else actually comes with the idea of just doing all the proteins in a month . Demi’s just agrees with the idea … takes the credit

2026-06-09 09:34:03

Lveran :

meanwhile they just discovered thousands of tiny proteins from areas of DNA that were thought dormant

2026-06-04 07:19:47

Concrete_Sleep :

The data baters?

2026-06-06 00:45:57

mrjukesU :

data-baters? 😏

2026-06-05 20:56:23

brian :

do memories next

2026-06-05 04:30:17

To see more videos from user @aidecipher, please go to the Tikwm homepage.

AI For People Tired of Noise

Open In TikTok:

Region: US

Thursday 04 June 2026 00:55:03 GMT

Music

Download

Comments

EMBER & YUKI :

Fernando M :

without ai this would have taken thousands of years

Carter :

And the first caveman to pick up a stick assumed he had mastered weaponry

brian :

I ran openfold on PlayStation 3 when it first came out

Dad Top Tips !!! :

Now tell it to apply that to predict what other intelligent species could possibly look like

Chimeras :

You can "predict" what you want, it doesn't mean it's active though.

Homosapienish :

Yes

pervywizard6 :

Instead of writing a novel, let's build the tower of Babel. oh thanks I needed that.

authorAndrew :

in that meeting someone else actually comes with the idea of just doing all the proteins in a month . Demi’s just agrees with the idea … takes the credit

Lveran :

meanwhile they just discovered thousands of tiny proteins from areas of DNA that were thought dormant

Concrete_Sleep :

The data baters?

mrjukesU :

data-baters? 😏

brian :

do memories next

Other Videos

About

Legal