How close are we to uploading our minds? - Michael S.A. Graziano

譯者: Clement Fu 審譯者: Zoe Chang

Imagine a future where nobody dies—

想像一下人類終將不死的一天

instead, our minds are uploaded to a digital world.

軀體不再，意識卻能上傳到數位世界

They might live on in a realistic, simulated environment with avatar bodies,

以類似阿凡達之軀活於模擬實境中

and could still call in and contribute to the biological world.

也能隨時回到實體世界做出貢獻

Mind uploading has powerful appeal—

意識上傳這件事有絕大吸引力

but what would it actually take to scan a person’s brain and upload their mind?

但實際應如何掃描上傳呢？

The main challenges are scanning a brain in enough detail to capture the mind

最大的挑戰在於如何掃描 捕捉腦中所有的細節

and perfectly recreating that detail artificially.

並將之完美地重現

But first, we have to know what to scan.

首先得決定掃描的內容

The human brain contains about 86 billion neurons,

人腦中有約 860 億個神經元

connected by at least a hundred trillion synapses.

由幾百兆個神經突觸連結著

The pattern of connectivity among the brain’s neurons,

腦部神經元的連結模式

that is, all of the neurons and all their connections to each other,

換句話說，就是整個腦部 神經細胞間線路連接的總合

is called the connectome.

稱為神經網路體

We haven’t yet mapped the connectome,

人類至今尚未能一窺全貌

and there’s also a lot more to neural signaling.

更別提腦神經的訊息傳遞了

There are hundreds, possibly thousands of different kinds of connections,

光是神經連結的方式 就有數百甚或上千種不同的神經突觸

or synapses.

Each functions in a slightly different way.

每種的功能都不盡相同

Some work faster, some slower.

有些運作得較快，有些較慢

Some grow or shrink rapidly in the process of learning;

有些在學習情境中急速變大縮小

some are more stable over time.

有些則隨著時間日趨穩定

And beyond the trillions of precise, 1-to-1 connections between neurons,

在這幾兆個神準緊密連結的神經元中

some neurons also spray out neurotransmitters

有些還能釋出神經傳導物質

that affect many other neurons at once.

在同一個時間內影響其他的神經元

All of these different kinds of interactions

這一切形形色色的神經元互動

would need to be mapped in order to copy a person’s mind.

需得被一一分析記錄下來 才能談論複製意識

There are also a lot of influences on neural signaling

另外有許多影響神經訊息傳導的因素

that are poorly understood or undiscovered.

至今我們仍所知有限或未能探究

To name just one example,

隨便舉個例子

patterns of activity between neurons

神經元之間的活動模式

are likely influenced by a type of cell called glia.

很容易受神經膠質細胞影響

Glia surround neurons and, according to some scientists,

環繞在神經元周圍的神經膠質細胞

may even outnumber them by as many as ten to one.

據科學家估計數量多達十比一

Glia were once thought to be purely for structural support,

過去以為膠質細胞 是神經元的結構支撐

and their functions are still poorly understood,

至今我們對其功能仍所知有限

but at least some of them can generate their own signals

現在我們已知其中一些能發出訊號

that influence information processing.

影響訊息處理

Our understanding of the brain isn’t good enough to determine

我們對人腦的認知仍不足以判斷

what we’d need to scan in order to replicate the mind,

掃描大腦哪些部位 才能百分百複製意識

but assuming our knowledge does advance to that point,

但即使我們所知已夠進展到那個階段

how would we scan it?

又該如何執行掃描？

Currently, we can accurately scan a living human brain

MRI 磁共振掃描是目前 最先進的非侵入式掃描

with resolutions of about half a millimeter

也是掃描活人腦部最精密的儀器

using our best non-invasive scanning method, MRI.

精密度高達 0.5 毫米解析度

To detect a synapse, we’ll need to scan at a resolution of about a micron—

但是要能掃描出神經突觸 解析度需得達到一微米

a thousandth of a millimeter.

亦即千分之一毫米

To distinguish the kind of synapse and precisely how strong each synapse is,

要能分辨神經突觸的種類 以及所發出信號的強弱

we’ll need even better resolution.

解析度還需要更高才行

MRI depends on powerful magnetic fields.

磁共振掃描靠的是磁場的強度

Scanning at the resolution required

要能掃描出每個突觸的細節

to determine the details of individual synapses

必須達到一定的解析度

would requires a field strength high enough to cook a person’s tissues.

所需的磁力足以煮熟人體組織細胞

So this kind of leap in resolution

因此要達到那樣的解析度

would require fundamentally new scanning technology.

我們需要全新的掃描技術

It would be more feasible to scan a dead brain using an electron microscope,

電子顯微鏡掃描死人的大腦或許可行

but even that technology is nowhere near good enough–

然而那技術還不夠好

and requires killing the subject first.

因為它對活體造成死亡威脅

Assuming we eventually understand the brain well enough to know what to scan

假設有那麼一天知道該掃描何處

and develop the technology to safely scan at that resolution,

也開發出高解析的安全掃描技術

the next challenge would be to recreate that information digitally.

還有下一個挑戰：資訊的數位重建

The main obstacles to doing so are computing power and storage space,

目前主要障礙 是電腦運算力和儲存空間

both of which are improving every year.

這兩方面每年都在進步

We’re actually much closer to attaining this technological capacity

因此我們離技術目標越來越近

than we are to understanding or scanning our own minds.

但離全面理解或掃描意識還遠得很呢

Artificial neural networks already run our internet search engines,

人工神經網路早已被運用在 架構網路搜尋引擎上

digital assistants, self-driving cars, Wall Street trading algorithms,

數位助理、自動駕駛汽車

華爾街交易運算及智慧型手機等等

and smart phones.

Nobody has yet built an artificial network with 86 billion neurons,

目前尚未出現高達 860 億個 神經元架構的人工網路

but as computing technology improves,

但隨著運算科技的進步

it may be possible to keep track of such massive data sets.

或許未來終能處理 如此龐大的數據資料

At every step in the scanning and uploading process,

無論掃描還是上傳

we’d have to be certain we were capturing all the necessary information accurately—

都須確保所有資訊 均能準確無誤地擷取

or there’s no telling what ruined version of a mind might emerge.

因為掃描毀壞的意識版本 會有何後果無人能預料

While mind uploading is theoretically possible,

儘管意識上傳理論上可行

we’re likely hundreds of years away

距離達到科學與技術那一天

from the technology and scientific understanding

似乎也尚有數百年

that would make it a reality.

才能實現

And that reality would come with ethical and philosophical considerations:

隨之而來的是道德與哲學的考量：

who would have access to mind uploading?

誰來掌控意識上傳？

What rights would be accorded to uploaded minds?

能被上傳的意識又能擁有哪些權力？

How could this technology be abused?

這個科技會否被濫用？

Even if we can eventually upload our minds,

即使真有意識上傳的一天

whether we should remains an open question.

到底該不該做仍舊沒有一個答案