TED演講者：Frances S. Chance / 弗朗西斯·S·錢斯

演講標(biāo)題：Are insect brains the secret to great AI? / 昆蟲大腦是偉大人工智能的秘密嗎？

內(nèi)容概要：Are insects the key to brain-inspired computing? Neuroscientist Frances S. Chance thinks so. In this buzzy talk, she shares examples of the incredible capabilities of insects -- like the dragonfly's deadly accurate hunting skills and the African dung beetle's superstrength -- and shows how untangling the mysterious web of neurons in their tiny brains could lead to breakthroughs in computers, AI and more.

昆蟲是大腦啟發(fā)計算的關(guān)鍵嗎？神經(jīng)科學(xué)家弗朗西斯·S·錢斯是這樣認(rèn)為的。在這場有趣的講座中，她分享了昆蟲不可思議的能力的例子——比如蜻蜓極精確狩獵技能和非洲蜣螂的超強(qiáng)力量——并展示了解開它們微小大腦中神秘的神經(jīng)元網(wǎng)絡(luò)如何導(dǎo)致計算機(jī)、人工智能等領(lǐng)域的突破。

*******************************************

【1】Creating intelligence on a computer.

在計算機(jī)上創(chuàng)建智能。

【2】This has been the Holy Grail for artificial intelligence for quite some time.

很長一段時間以來,這一直是 人工智能的圣杯。

【3】But how do we get there?

但我們?nèi)绾蔚竭_(dá)那里？

【4】So we view ourselves as highly intelligent beings.

我們認(rèn)為自己是高度智慧的人。

【5】So it's logical to study our own brains, the substrate of our cognition, for creating artificial intelligence.

因此,研究我們自己的大腦, 我們認(rèn)知的基礎(chǔ), 來創(chuàng)造人工智能是合乎邏輯的。

【6】Imagine if we could replicate how our own brains work on a computer.

想象一下,如果我們可以在計算機(jī)上 復(fù)制我們自己的大腦是如何工作的。

【7】But now consider the journey that would be required.

但現(xiàn)在考慮一下所需的過程。

【8】The human brain contains 86 billion neurons.

人類大腦包含 860 億個神經(jīng)元。

【9】Each is constantly communicating with thousands of others, and each has individual characteristics of its own.

每個人都在不斷地 與成千上萬的人交流, 每個人都有自己的特點(diǎn)。

【10】Capturing the human brain on a computer may simply be too big and too complex a problem to tackle with the technology and the knowledge that we have today.

在計算機(jī)上捕獲人腦智慧 可能的確是一個太大、 太復(fù)雜的問題, 無法用我們今天的技術(shù)和知識來解決。

【11】I believe that we can capture a brain on a computer, but we have to start smaller.

我相信我們可以在計算機(jī)上捕獲智慧, 但我們必須從更小的地方開始。

【12】Much smaller.

小得多。

【13】These insects have three of the most fascinating brains in the world to me.

對我來說,這些昆蟲有三個 世界上最迷人的大腦。

【14】While they do not possess human-level intelligence, each is remarkable at a particular task.

雖然它們不具備人類水平的智力, 但每一個都在特定活動中表現(xiàn)出色。

【15】Think of them as highly trained specialists.

將他們視為訓(xùn)練有素的專家。

【16】African dung beetles are really good at rolling large balls in straight lines.

非洲屎殼郎真的很擅長 在直線上滾動大球。

【17】Now, if you've ever made a snowman, you know that rolling a large ball is not easy.

如果你曾經(jīng)堆過雪人, 你就知道滾一個大球并不容易。

【18】Now picture trying to make that snowman when the ball of snow is as big as you are and you're standing on your head.

現(xiàn)在想象一下堆雪人 當(dāng)雪球和你一樣大時, 你倒立著。

【19】Sahara desert ants are navigation specialists.

撒哈拉沙漠螞蟻是導(dǎo)航專家。

【20】They might have to wander a considerable distance to forage for food.

他們可能要走很遠(yuǎn)的路才能覓食。

【21】But once they do find sustenance, they know how to calculate the straightest path home.

但一旦他們找到了食物, 他們就知道如何計算回家的最直路徑。

【22】And the dragonfly is a hunting specialist.

而蜻蜓是狩獵專家。

【23】In the wild, dragonflies capture approximately 95 percent of the prey they choose to go after.

在野外,蜻蜓捕獲了大約 95% 的 它們選擇的獵物。

【24】These insects are so good at their specialties that neuroscientists such as myself study them as model systems to understand how animal nervous systems solve particular problems.

這些昆蟲非常擅長它們的專業(yè), 以至于像我這樣的神經(jīng)科學(xué)家 將它們作為模型系統(tǒng)來研究, 以了解動物神經(jīng)系統(tǒng) 是如何解決特定的問題。

【25】And in my own research, I study brains to bring these solutions, the best that biology has to offer, to computers.

在我的研究中, 我研究大腦, 以將這些生物所能提供的 最好的解決方案引入計算機(jī)。

【26】So consider the dragonfly brain.

想一下蜻蜓的大腦。

【27】It has only on the order of one million neurons.

它只有大約 100 萬個神經(jīng)元。

【28】Now, it's still not easy to unravel a circuit of even one million neurons.

現(xiàn)在,要解開一個哪怕有一百萬個 神經(jīng)元的回路仍然不容易。

【29】But given the choice between trying to tease apart the one-million-neuron brain versus the 86-billion-neuron brain, which would you choose to try first?

但是如果要在 嘗試梳理 100 萬個神經(jīng)元大腦 和 860 億個神經(jīng)元大腦之間做出選擇, 你會選擇先嘗試哪一個？

【30】When studying these smaller insect brains, the immediate goal is not human intelligence.

當(dāng)研究這些較小的昆蟲大腦時, 當(dāng)前的目標(biāo)不是人類的智力。

【31】We study these brains for what the insects do well.

我們研究這些大腦是為了 了解昆蟲做得好的地方。

【32】And in the case of the dragonfly, that's interception.

就蜻蜓而言,那就是攔截。

【33】So when dragonflies are hunting, they do more than just fly straight at the prey.

因此,當(dāng)蜻蜓捕食時, 它們所做的不僅僅是直接飛向獵物。

【34】They fly in such a way that they will intercept it.

它們以這樣的方式飛行,以攔截它。

【35】They aim for where the prey is going to be.

它們瞄準(zhǔn)獵物將要到達(dá)的地方。

【36】Much like a soccer player, running to intercept a pass.

就像足球運(yùn)動員, 跑去攔截傳球。

【37】To do this correctly, dragonflies need to perform what is known as a coordinate transformation, going from the eye's frame of reference, or what the dragonfly sees, to the body's frame of reference, or how the dragonfly needs to turn its body to intercept.

為了正確地做到這一點(diǎn), 蜻蜓需要進(jìn)行所謂的坐標(biāo)變換, 從眼睛的參照系或蜻蜓看到的東西, 到身體的參照系, 或者蜻蜓需要如何轉(zhuǎn)動身體進(jìn)行攔截。

【38】Coordinate transformations are a basic calculation that animals need to perform to interact with the world.

坐標(biāo)變換是動物 與世界互動所需要進(jìn)行的基本計算。

【39】We do them instinctively every time we reach for something.

我們每次伸手拿東西的時候 都會本能地做這些計算。

【40】When I reach for an object straight in front of me, my arm takes a very different trajectory than if I turn my head, look at that same object when it is off to one side and reach for it there.

當(dāng)我伸手去拿我面前的一個物體時, 我的手臂的運(yùn)動軌跡 和我轉(zhuǎn)頭看向一邊的同一物體時 完全不同。

【41】In both cases, my eyes see the same image of that object, but my brain is sending my arm on a very different trajectory based on the position of my neck.

在這兩種情況下,我的眼睛 看到的都是同一物體的圖像, 但我的大腦根據(jù)我脖子的位置 將我的手臂送上一個 非常不同的軌跡。

【42】And dragonflies are fast.

蜻蜓很快。

【43】This means they calculate fast.

這意味著他們計算得很快。

【44】The latency, or the time it takes for a dragonfly to respond once it sees the prey turn, is about 50 milliseconds.

延遲,即蜻蜓在看到獵物轉(zhuǎn)向后 做出反應(yīng)所需的時間, 大約是 50 毫秒。

【45】This latency is remarkable.

這種延遲是很了不起的。

【46】For one thing, it's only half the time of a human eye blink.

一方面,這只是人類眨眼時間的一半。

【47】But for another thing, it suggests that dragonflies capture how to intercept in only relatively or surprisingly few computational steps.

但另一方面, 它表明蜻蜓僅通過相對的 或驚人的極少計算步驟 即可體現(xiàn)出如何進(jìn)行攔截。

【48】So in the brain, a computational step is a single neuron or a layer of neurons working in parallel.

所以在大腦中, 計算步驟是單個神經(jīng)元 或一層神經(jīng)元并行工作。

【49】It takes a single neuron about 10 milliseconds to add up all its inputs and respond.

單個神經(jīng)元需要大約 10 毫秒 才能將其所有輸入相加并做出反應(yīng)。

【50】The 50-millisecond response time means that once the dragonfly sees its prey turn, there's only time for maybe four of these computational steps or four layers of neurons, working in sequence, one after the other, to calculate how the dragonfly needs to turn.

50毫秒的響應(yīng)時間意味著, 一旦蜻蜓看到它的獵物轉(zhuǎn)向, 可能只有四個計算步驟 或四層神經(jīng)元依次工作的時間, 一個接一個, 來計算蜻蜓需要如何轉(zhuǎn)向。

【51】In other words, if I want to study how the dragonfly does coordinate transformations, the neural circuit that I need to understand, the neural circuit that I need to study, can have at most four layers of neurons.

換句話說,如果我想研究 蜻蜓如何進(jìn)行坐標(biāo)變換, 我需要了解神經(jīng)回路, 我需要研究神經(jīng)回路, 最多可以有四層神經(jīng)元。

【52】Each layer may have many neurons, but this is a small neural circuit.

每一層可能有許多神經(jīng)元, 但這是一個小的神經(jīng)回路。

【53】Small enough that we can identify it and study it with the tools that are available today.

小到我們可以用 今天的工具來識別它和研究它。

【54】And this is what I'm trying to do.

這就是我要做的。

【55】I have built a model of what I believe is the neural circuit that calculates how the dragonfly should turn.

我已經(jīng)建立了一個我認(rèn)為是計算 蜻蜓應(yīng)該如何轉(zhuǎn)向的神經(jīng)回路的模型。

【56】And here is the cool result.

這是一個很酷的結(jié)果。

【57】In the model, dragonflies do coordinate transformations in only one computational step, one layer of neurons.

在該模型中, 蜻蜓只用一個計算步驟, 即一個神經(jīng)元層 來做坐標(biāo)轉(zhuǎn)換。

【58】This is something we can test and understand.

這是我們可以測試和理解的。

【59】In a computer simulation, I can predict the activities of individual neurons while the dragonfly is hunting.

在計算機(jī)模擬中, 我可以預(yù)測蜻蜓狩獵時 單個神經(jīng)元的活動。

【60】For example, here I am predicting the action potentials, or the spikes, that are fired by one of these neurons when the dragonfly sees the prey move.

例如,我在這里預(yù)測 當(dāng)蜻蜓看到獵物移動時, 其中一個神經(jīng)元 發(fā)射了動作電位或脈沖。

【61】To test the model, my collaborators and I are now comparing these predicted neural responses with responses of neurons recorded in living dragonfly brains.

為了測試這個模型, 我和我的合作者 現(xiàn)在正在將這些預(yù)測的神經(jīng)反應(yīng) 與活體蜻蜓大腦中 記錄的神經(jīng)元反應(yīng)進(jìn)行比較。

【62】These are ongoing experiments in which we put living dragonflies in virtual reality.

這些是正在進(jìn)行的實(shí)驗(yàn), 我們將活體蜻蜓放在虛擬現(xiàn)實(shí)中。

【63】Now, it's not practical to put VR goggles on a dragonfly.

現(xiàn)在,給蜻蜓戴上 VR 護(hù)目鏡是不現(xiàn)實(shí)的。

【64】So instead, we show movies of moving targets to the dragonfly, while an electrode records activity patterns of individual neurons in the brain.

因此,我們改為向蜻蜓 播放移動目標(biāo)的電影, 同時電極記錄大腦中單個神經(jīng)元的 活動模式。

【65】Yeah, he likes the movies.

是的,他喜歡電影。

【66】If the responses that we record in the brain match those predicted by the model, we will have identified which neurons are responsible for coordinate transformations.

如果我們在大腦中記錄的反應(yīng) 與模型預(yù)測的反應(yīng)相匹配, 我們就會確定哪些神經(jīng)元 負(fù)責(zé)坐標(biāo)轉(zhuǎn)換。

【67】The next step will be to understand the specifics of how these neurons work together to do the calculation.

下一步將是了解這些神經(jīng)元 如何協(xié)同工作進(jìn)行計算的細(xì)節(jié)。

【68】But this is how we begin to understand how brains do basic or primitive calculations.

但這就是我們開始了解大腦 是如何進(jìn)行基本 或原始的計算。

【69】Calculations that I regard as building blocks for more complex functions, not only for interception but also for cognition.

計算,我將其視為更復(fù)雜功能的構(gòu)件, 不僅用于攔截, 還用于認(rèn)知。

【70】The way that these neurons compute may be different from anything that exists on a computer today.

這些神經(jīng)元的計算方式可能不同于 當(dāng)今計算機(jī)上存在的任何東西。

【71】And the goal of this work is to do more than just write code that replicates the activity patterns of neurons.

這項(xiàng)工作的目標(biāo)不僅僅是 編寫復(fù)制神經(jīng)元活動模式的代碼。

【72】We aim to build a computer chip that not only does the same things as biological brains but does them in the same way as biological brains.

我們的目標(biāo)是制造一種計算機(jī)芯片, 它不僅可以做 與生物大腦相同的事情, 而且可以用與生物大腦同樣的方式 來做這些事情。

【73】This could lead to drones driven by computers the same size of the dragonfly's brain that captures some targets and avoid others.

這可能會導(dǎo)致由計算機(jī)驅(qū)動的無人機(jī), 其大小與蜻蜓的大腦相同, 捕獲一些目標(biāo)并避開其他目標(biāo)。

【74】Personally, I'm hoping for a small army of these to defend my backyard from mosquitoes in the summer.

就我個人而言, 我希望有一小群這樣的無人機(jī) 在夏天保護(hù)我的后院不受蚊子騷擾。

【75】The GPS on your phone could be replaced by a new navigation device based on dung beetles or ants that could guide you to the straight or the easy path home.

你手機(jī)上的 GPS 可能會被一種 基于蜣螂或螞蟻的新型導(dǎo)航設(shè)備所取代, 它可以引導(dǎo)你走直路或容易回家的路。

【76】And what would the power requirements of these devices be like?

那么這些設(shè)備的功率要求是怎樣的呢?

【77】As small as it is - Or, sorry -- as large as it is, the human brain is estimated to have the same power requirements as a 20-watt light bulb.

盡管它很小, 或者說,對不起,盡管它很大, 據(jù)估計,人腦的功率需求 與 20 瓦的燈泡相同。

【78】Imagine if all brain-inspired computers had the same extremely low-power requirements.

想象一下, 如果所有受大腦啟發(fā)的計算機(jī) 都具有相同的極低功耗要求。

【79】Your smartphone or your smartwatch probably needs charging every day.

你的智能手機(jī)或智能手表 可能每天都需要充電。

【80】Your new brain-inspired device might only need charging every few months, or maybe even every few years.

你的新大腦啟發(fā)設(shè)備 可能只需要每隔幾個月, 甚至幾年充電一次。

【81】The famous physicist, Richard Feynman, once said, "What I cannot create, I do not understand."

著名物理學(xué)家理查德.費(fèi)曼曾說: “我不能創(chuàng)造的東西,我就不了解?！?/p>

【82】What I see in insect nervous systems is an opportunity to understand brains through the creation of computers that work as brains do.

我在昆蟲神經(jīng)系統(tǒng)中看到的 是一個通過創(chuàng)造 與大腦一樣工作的計算機(jī) 來了解大腦的機(jī)會。

【83】And creation of these computers will not just be for knowledge.

而這些計算機(jī)的創(chuàng)造 將不僅僅是為了認(rèn)知。

【84】There's potential for real impact on your devices, your vehicles, maybe even artificial intelligences.

有可能對你的設(shè)備、車輛 甚至是人工智能產(chǎn)生真正的影響。

【85】So next time you see an insect, consider that these tiny brains can lead to remarkable computers.

所以,下次你看到一只昆蟲時, 想想看,這些微小的大腦 可以發(fā)展出卓越的計算機(jī)。

【86】And think of the potential that they offer us for the future.

想想它們?yōu)槲覀兊奈磥硖峁┑臐摿Α?/p>

【87】Thank you.

謝謝。