We dey live for three-dimensional space.
Inside dis space, we dey see three-dimensional space based on wetin our eye see, wey be just two-dimensional picture.
Dis mean say our brain get one picture of three-dimensional space, and we dey use dis picture to understand two-dimensional visual information.
I believe say if we apply dis idea, humans fit even see four-dimensional space. Even though we no fit create four-dimensional space or objects for di real world,
we fit use computer to simulate four-dimensional space and objects. If we now map dis simulated four-dimensional space onto a two-dimensional flat surface, humans fit see and understand di information.
Then, if humans learn how dis four-dimensional space and objects dey behave and look, dem go eventually fit create a four-dimensional space for inside their mind.
But, dis one na just possibility, and e go probably take plenty time to learn am.
Also, even if person come fit see four-dimensional space, almost no situation go dey where dem fit use dis skill.
How AI Dey See Four Dimensions
On di other hand, AI fit do di same thing. And e pass dat, AI fit even use dis four-dimensional way of seein' space.
For example, if AI get four-dimensional spatial perception, e fit draw and understand four-dimensional graphs.
Humans only fit see visual information on a two-dimensional flat surface at once. Because of dis, even if dem draw a three-dimensional graph and understand am by inverse mapping, some hidden parts go still dey wey eye no go see.
Even for three-dimensional graph, plenty parts go disappear, but for four-dimensional graph, more data go hide.
Although you fit rotate di graph to show di hidden parts, dis one no go achieve di aim of understandin' data sharp sharp at a glance.
But for AI, e no need to dey tied to two-dimensional flat visual information. You fit make AI get three-dimensional or four-dimensional virtual spatial vision and train am.
Dis one go allow AI to understand three-dimensional and four-dimensional graphs for a dimension-native, wide view, without data hidin' or needin' rotation.
And e no just stop at four dimensions; logically, dem fit increase dimensions infinitely to five, ten, twenty, and even more.
How To Understand Plenty-Dimension Graphs
If you fit see graphs from far and understand them well, e go make am possible to do things like check trends across different dimensions. You go also fit compare sizes and understand ratios easily.
E also dey make am possible to analyze data patterns, like data wey be similar or identical. Apart from that, e fit help find orderliness and rules.
Dis one pass just matching patterns for plenty-dimension data, wey normal AI dey sabi do well. E dey make am possible to understand data better.
For example, even if di same patterns dey for different dimension combinations, simple plenty-dimension pattern matching go probably struggle to find them.
But, if you get plenty-dimension vision, if di shapes similar, you go quick identify them, even if dem dey for different dimension combinations.
Again, apart from just using di dimension axes wey come with di input data, you fit also explore dimension structures wey go make am easier to understand di data. You fit do dis by expanding or contracting specific axes, changing them to logarithms, or mapping plenty axes to di same number of different axes without reducing their dimensions.
So, by training di skill of plenty-dimension vision, e go be possible to understand data structures from far—wey be hard for both humans and normal AI—and dis go open door for us to discover new knowledge and laws from them.
Makin' Paradigm Innovation Faster
Di power to understand plenty-dimension data directly, without havin' to simplify am to fewer dimensions, dey show say plenty good things fit happen.
For example, dem invent di heliocentric theory so dat astronomical data wey dem observe fit enter mathematical formulas wey easy to understand. Di geocentric idea, wey talk say sun dey rotate Earth, no fit arrange di observation data into formulas wey pipu go understand. Na dis one lead to di invention of heliocentrism.
But, if dem fit understand astronomical observation data directly without changin' di dimensions, dem for don discover laws like heliocentrism long time ago.
Like dat too, scientific inventions like di theory of relativity and quantum mechanics for don come out much earlier if dem fit understand plenty-dimension data well well for its original dimensions.
Dis one mean say dimension-native AI fit make paradigm innovation faster, wey go lead to di discovery of many theories and laws wey humanity never know before.
Conclusion
AI wey dem train to sabi plenty-dimension space well well, by using its plenty-dimension way of thinking about space—wey pass wetin human fit copy—fit quick make di scope of paradigms for science and school wide.
Paradigms dey increase pass just changing from one to anoda. Even if dem invent new paradigms, we no necessarily need to catch up with dem.
Of course, AI go probably explain paradigms wey dem discover for complex dimensions by changing dem to smaller dimensions for a way wey easy for us to understand.
Still, paradigms wey get too many dimensions fit still dey hard for human to understand completely. Also, e go probably be impossible to understand all di plenty paradigms wey don expand.
For such a situation, we fit come see ourself dey live surrounded by products and systems wey dey work well, even if we no fully understand how dem dey work.
As an engineer, I no really want to imagine such a situation, but for plenty pipo, e fit no be too different from how things dey today.