《经济学人》AI造物时代到来？从代码重构到凭空制造生命 [C

共 1 集

剧集目录

集	标题	时长
P1	《经济学人》AI造物时代到来？从代码重构到凭空制造生命 [C	14:04

P1 · 《经济学人》AI造物时代到来？从代码重构到凭空制造生命 [C1] p01 原声 (P1)

《经济学人》AI造物时代到来？从代码重构到凭空制造生命 [C1] p01 原声 (P1)

发布于 2026-10-27 · 时长 14:04

字幕摘录

时间	英文	中文
0:18	20 years ago the economist ran a story about a factory that could make dragons.	20年前,经济学家运行了一个关于工厂可以制造龙。
0:23	It was	这是
0:23	an April Fool based on the idea that living cells could be modelled accurately inside	April Fool基于的理念是,活着的细胞可以准确地模拟在内
0:29	computers.	电脑。
0:30	At the time that idea was a fantasy, but the flourishing of artificial intelligence	当时这个想法是幻想但人工智能的繁荣
0:35	in the intervening decades appears to have brought it closer to reality.	在中间几十年中,它似乎更接近现实。
0:41	A sign of progress	进步的迹象
0:41	came on October 7 when the Virtual Biology Initiative, or VBI, a project to collect the	10月7日,虚拟生物学倡议或VBI,一个收集
0:48	data needed to create predictive models of how cells behave and change, announced that	为创建细胞行为和变化预测模型所需的数据,宣布
0:55	it had received commitments worth $800 million since its launch in April.	自4月启动以来,它收到了价值8亿美元的承付款。

展开字幕全文（205 条）

序号	英文	中文
1	20 years ago the economist ran a story about a factory that could make dragons.	20年前,经济学家运行了一个关于工厂可以制造龙。
2	It was	这是
3	an April Fool based on the idea that living cells could be modelled accurately inside	April Fool基于的理念是,活着的细胞可以准确地模拟在内
4	computers.	电脑。
5	At the time that idea was a fantasy, but the flourishing of artificial intelligence	当时这个想法是幻想但人工智能的繁荣
6	in the intervening decades appears to have brought it closer to reality.	在中间几十年中,它似乎更接近现实。
7	A sign of progress	进步的迹象
8	came on October 7 when the Virtual Biology Initiative, or VBI, a project to collect the	10月7日,虚拟生物学倡议或VBI,一个收集
9	data needed to create predictive models of how cells behave and change, announced that	为创建细胞行为和变化预测模型所需的数据,宣布
10	it had received commitments worth $800 million since its launch in April.	自4月启动以来,它收到了价值8亿美元的承付款。
11	VBI的目标
12	are medical.	是医疗。
13	Predictive models will let researchers design more effective drugs that are more	预测模型将使研究人员能够设计更有效的药物
14	likely to pass successfully through clinical trials.	有可能成功通过临床试验。
15	But if that works, future generations	但是如果成功的话,子孙后代
16	of models might be used to design modified versions of existing organisms, such as crop	可使用模型设计现有生物体的修改版本,例如作物
17	and perhaps, eventually, entirely new living things.	并且,也许,最终, 全新的生物。
18	The technology that has made such	创造这些技术的技术
19	advances conceivable is deep learning, wherein large neural networks are trained to tackle	可以想象,进步是深入的学习,在这种学习中,大型神经网络接受培训,以应对
20	complex problems.	复杂的问题。
21	It is this technique that underpins the foundation models that power	正是这种技术支撑了基础模型的力量
22	由Anthropic、Google和OpenAI等公司销售的聊天机器人。
23	Two such models, the long	两个这样的模型,长
24	acids and proteins, are no different from the chains of words in human writing.	酸和蛋白质,与人类写作中的词链无异.
25	Letting	让开
26	foundation models loose on them was thus an obvious idea, and it is proving extraordinarily	因此,基础模型的松散是一个显而易见的想法,它正在证明特别
27	productive.	生产。
28	In the words of Adrian Wolfson, the business brain behind GENERO, a biostartup	用阿德里安·沃尔夫森的话来说,GENERO背后的商业大脑, 一个生物启动
29	in San Diego, biology is now computable, predictable, and therefore designable.	在圣地亚哥,生物学现在可以计算,可以预测,因此可以设计.
30	It is also instantiable.	它也是即时的。
31	例如EVO2,一个基础模型由Brian He of the
32	ARC Institute in Palo Alto, can analyze DNA segments up to one million base pairs long,	Palo Alto的ARC研究所可以分析DNA片段,可长达100万个碱基对,
33	the size of a small bacterial genome.	小型细菌基因组的大小。
34	By studying the order of chemical bases paired together	通过研究化学基础的组合顺序
35	in that molecule's double helix, it can predict how changes to that order might affect its	在分子的双螺旋中,它可以预测改变该顺序会如何影响它
36	organism.	有机体。
37	It is able to do this because it was trained on DNA sequences collected from over	它能够做到这一点,因为它接受了关于从以上收集的DNA序列的培训。
38	100,000 species in the way that chatbots are trained on text.	10万个物种以方式聊天员接受文字培训。
39	In January, Dr. Wolfson's	一月,沃尔夫森医生
40	加州理工学院Kai-Hung Wong和Noah Robinson的科学伙伴
41	unleashed on the world a high-fidelity, high-efficiency DNA assembly system called	在世界上释放出一种高品质,高效率的DNA组装系统
42	Sidewinder.	侧风者.
43	Sidewinder can piece long stretches of DNA together, currently up to 12,500 base pairs,	侧风机可以把长长的DNA拉在一起, 目前高达12500个碱基对,
44	with high accuracy.	准确度很高。
45	That is enough by itself to synthesize the genomes of small viruses.	这本身就足以合成小病毒的基因组。
46	It also simplifies the construction of yet longer stretches, such as bacterial genomes and small	它还简化了更长的拉伸的构造,例如细菌基因组和小的
47	artificial chromosomes.	人工染色体.
48	与EVO2合并,Dr.
49	He is another of GENERO's founders,	他是GENERO的创始人之一
50	the firm now has the tools needed for Dr. Wolfson to put his money where his mouth is.	公司现在有Wolfson医生所需的工具把钱放在嘴边
51	He hopes to do so by synthesizing, among other things, an artificial wheat chromosome carrying	他希望通过合成人造小麦染色体来做到这一点。
52	genes for heat and drought resistance.	用于抗热抗旱的基因.
53	EVO2 is by no means the only genomic foundation model out	EVO2绝非唯一的基因组基础模型
54	there.	这里
55	DeepMind,Google的AI实验室,创建了一个名为AlphaGenome的实验室.
56	It is more of a specialist	更像是专家
57	比EVO2还多.
58	AlphaGenome was trained almost exclusively on human genomes, as was Genos,	AlphaGenome和Genos一样,几乎完全接受过人类基因组培训,
59	an offering from BGI Research in Shenzhen, China.	中国深圳BGI研究的献礼.
60	Deep Learning models are available,	有深层学习模式,
61	too, for those studying particular groups of non-human species.	对于研究特定群体非人类物种的人来说也是如此。
62	Plants, fungi, bacteria,	植物,真菌,细菌,
63	viruses and vertebrate animals are all served in this way.	病毒和脊椎动物都以这种方式服役.
64	反之,Basecamp研究,
65	a company in London, is casting the net wider.	伦敦的一家公司正在扩大网络范围
66	Its model is trained on samples that attempt to	其型号在试图
67	capture all the DNA in, say, a water sample from a particular place, in order to study,	从某个地方收集所有DNA样本以便研究
68	with a view to creating new antibiotics, the chemical warfare going on between the microbes	为了制造新的抗生素微生物之间的化学战
69	therein.	这里。
70	And that is only genomics.	而这只是基因组学。
71	Deep Learning models of various sorts can also handle other	各种深层学习模型也可以处理其他
72	biological molecules, especially proteins.	生物分子,特别是蛋白质.
73	They are able to work out how natural proteins,	他们能够研究出自然蛋白质是如何形成的,
74	which are long chains of chemical links called amino acids, fold up, which is crucial to	长链的化学链叫做氨基酸,折叠起来,这对
75	understanding how they work.	理解他们是如何工作的。
76	They can also predict what changing particular amino acids will do,	他们也可以预测改变特定的氨基酸会做什么,
77	and even design proteins from scratch, with particular purposes in mind, from better	甚至设计蛋白质从零开始, 特别的目的,从更好的
78	antidotes to snake venom, to bespoke enzymes for novel chemical reactions.	解毒剂用于蛇毒,为新颖的化学反应代言酶.
79	Deep Mind also	深思熟虑
80	其中有一个叫做AlphaFold3,它的副作用是Isomorphic Labs,它拥有微小的
81	mission statement, Solve All Disease, plans to use both that and AlphaGeno in drug discovery	任务说明,"解决所有疾病",计划用这个和AlphaGeno两种方法进行药物发现.
82	partnerships with three larger, longer established pharma firms, Novartis, Eli Lilly and Johnson &	与三个规模较大、历史较长的制药公司、诺华公司、Eli Lilly公司和Johnson &
83	Provide the Targets.	提供目标。
84	Isomorphic将使用它的AI推荐药物候选人.
85	The pharma firms will	制药公司会
86	then test its suggestions experimentally.	然后试验它的建议。
87	Other established pharma companies are doing similar	其他的制药公司也在做类似的事情
88	AI work in-house.	AI内部工作.
89	Roche、Novo、GSK和AstraZeneca要么都拥有,要么有股份,
90	supercomputers for that purpose.	超级电脑
91	Indeed, Lilly itself has hedged its bets by building one	事实上,莉莉自己也通过建造一个
92	离开Isomorphic。
93	This will apply the deep learning approach to data from the	这将采用深入的学习方法来收集数据。
94	company's 150-year-old archive, which is replete with unpublished failures as well as published	公司150年的档案充满了未出版的失败和出版
95	successes.	成绩。
96	That should yield insights unavailable to those relying solely on the published scientific	对于那些完全依赖已发表的科学论文的人来说,这应产生无法获得的见解。
97	literature with its bias towards what worked.	文学作品对什么是有效的。
98	In the field of agriculture, meanwhile, AI is	同时,在农业领域,大赦国际认为,
99	being brought to bear on several hitherto intractable problems.	在几个迄今难以解决的问题上
100	One such is polyploidy,	其中一种是多聚体,
101	a phenomenon common in plants in which an organism has multiple sets of similar chromosomes.	一种常见于植物中的现象,在植物中,生物体有多种相似的染色体.
102	Several	几个
103	important crops, including wheat, cotton and potatoes, are polyploid.	包括小麦、棉花和马铃薯在内的重要作物是多聚类作物。
104	This makes it hard to	这很难
105	determine which genes are responsible for what effects, and thus know where to make advantageous	确定哪些基因对何种影响负责,从而知道在何处有利
106	genetic edits.	基因编辑.
107	Appropriate AI can sort out what's what.	适当的AI可以解决什么是什么。
108	It can also help make sense of plant immune	也能帮助理解植物免疫
109	systems, which have evolved independently of those in animals, but are similarly complex.	系统与动物系统独立发展,但同样复杂。
110	That should allow them to be tweaked to pharma's advantage.	这应该可以让他们被调整成药材的优势。
111	And there are even attempts to use AI	甚至有人试图使用AI
112	to tinker with the fundamentals of photosynthesis, to try to make it more efficient.	努力提高光合作用的效率
113	A long shot,	一个很长的机会,
114	given that Evil Ocean has had several billion years to optimize it, but one that would be	邪恶的海洋有数十亿年的时间来优化它, 但其中之一将是
115	incalculable value if it could be pulled off to increase yields.	如果能够提高产量,则其价值不可计算。
116	The big prize, though,	不过大奖
117	will be to combine the approaches to biology enabled by AI to devise computer models of how	将结合人工智能所允许的生物学方法,设计计算机模型,说明如何利用这些方法。
118	cells work.	细胞工作。
119	The VBI is a step towards this.	VBI是朝这个方向迈出的一步.
120	The initiative itself is a coalition of public,	该举措本身是一个公众联盟,
121	academic and private research organizations, led by Biohub, an organization in San Francisco,	学术和私人研究组织,由旧金山的一个组织Biohub领导,
122	由梅塔的老板马克·扎克伯格(Mark Zuckerberg)和妻子陈普里西拉(Priscilla Chan)所建立.
123	深明,Isomorphic实验室,
124	and the Arc Institute are among its members.	Arc研究所是其成员之一。
125	The aim is to gather, in AI-friendly format,	目的是以方便AI的形式收集:
126	the vast amounts of new biological data needed to create predictive models.	创造预测模型需要大量新的生物数据。
127	Though these will	虽然这些会
128	not replicate all aspects of cellular behavior, they will respond like real cells to simulated	不复制细胞行为的所有方面,他们会像真细胞一样反应模拟
129	disease and treatment.	疾病和治疗。
130	That should greatly speed up medical research and drug testing.	这将大大加快医学研究和药物测试。
131	Virtual cells with different genomes, for example, might indicate which are likely to respond well to	例如,具有不同基因组的虚拟细胞可能表明哪些细胞很可能对基因组产生良好反应。
132	a drug even before clinical trials start.	甚至在临床试验开始之前就已经是药物了
133	That would make such trials more of a formality,	这将使这种审判更加正式,
134	rather than the nail-biting, years-long exercise they currently are.	而不是咬钉, 多年的锻炼他们现在。
135	To understand enough biology	要了解足够的生物学
136	to make such a virtual cell will require looking at and building models to handle	制作这样的虚拟单元格需要查看和建立模型来处理
137	a lot more than just the composition of individual DNA and protein molecules.	不仅仅是单个DNA和蛋白质分子的构成
138	These new models will	这些新模式将
139	need to know where those molecules are, how they interact with each other, and with the other	需要知道这些分子在哪里, 它们是如何相互作用的, 以及它们是如何相互作用的
140	molecules of biochemistry such as lipids and sugars, and how this whole network responds	脂类和糖类等生物化学分子以及整个网络的反应
141	to signaling molecules such as hormones.	以信号分子如激素。
142	The VBI will sponsor the development of tools to do this,	VBI将赞助开发这方面的工具,
143	such as cryo-electron tomography, which can resolve the insides of a cell at close to atomic detail.	例如低温电极成像法,它能以接近原子细节的方式解决细胞的内在.
144	Meanwhile, in a collaboration with Biohub, also announced on October 7th,	同时,与Biohub合作,10月7日也宣布,
145	America's Department of Energy, or DOE, and National Institutes of Health, or NIH,	美国能源部、DOE、国家卫生研究所、NIH
146	will chip in too.	也会被炸成碎片
147	The NIH will contribute $500 million worth of already collected data	NIH将捐助价值5亿美元已经收集的数据
148	that Biohub will attempt to standardise, and the DOE will spend a similar sum on computing	Biohub将试图实现标准化,而DOE将在计算上花费类似的金额
149	and other resources needed for the endeavour.	以及这项工作所需的其他资源。
150	Virtual cells will still be just computer models.	虚拟细胞仍将只是计算机模型.
151	Synthetic biology's masterpiece, in the words' original sense of a work created	合成生物学的杰作以文字的原创意义来说是创作的作品
152	to prove mastery, will be to create completely novel organisms from scratch.	将创造出全新的生物
153	Depending on how you define the words organisms and novel, this may already have happened.	取决于你如何定义生物体和新事物, 这可能已经发生了。
154	In September 2025, Dr.	2025年9月,获博士.
155	He and some of his colleagues at the ARC Institute announced	他和他的一些同事在ARC研究所宣布
156	that they had used EVO2 and its predecessor EVO1 to design and synthesise 16 viable viruses	他们使用EVO2及其前身EVO1来设计和合成16种可行的病毒
157	based on, but significantly different from, a bacteriophage called Phi X-174,	基于一种叫Phi X-174的细菌,但明显不同。
158	that is one of virology's workhorses.	那是病毒学的功劳之一
159	Bacteriophages, as their name suggests, prey on	如其名字所示,细菌是猎物
160	bacteria.	细菌。
161	Intriguingly, some of Dr.	有趣的是,一些医生。
162	He's novelties had higher fitness than their wild equivalents.	他的新奇人物的健身能力比他们的野生同类要高.
163	Cocktails of them proved able to overcome the normal resistance to viruses shown by three	他们的鸡尾酒能够克服正常的抗病毒能力由三个人所显示
164	strains of E. coli, a gut bacterium that also has workhorse status.	E. 大肠杆菌的菌株,一种也有工作马的肠道杆菌。
165	This is of particular interest	这特别值得关注
166	because custom-built phages might have medical applications.	因为定制的火腿可能有医疗用途
167	So-called phage therapies,	所谓的香草疗法,
168	in which hostile bacteria are kept under control by viruses rather than chemical antibiotics,	在其中敌对细菌由病毒而不是化学抗生素控制,
169	were once widely regarded as eccentric, but are now under active consideration.	曾被广泛认为是古怪的,但目前正在积极考虑。
170	Back at Genero, no one could accuse Dr. Wilson of thinking small.	回到Genero 没人能指责Wilson医生小题大做
171	He hopes the combination of	他希望结合
172	biological foundation models and cheap large-scale DNA synthesis will enable the exploration of the	生物基础模型和廉价的大型DNA合成将有助于探索
173	myriad possibilities offers that natural selection has not yet discovered what he calls the whole	有很多可能性自然选择还没有发现他所谓的整体
174	DNA空间
175	This would presumably start with creatures that, though novel, still had,	这大概会从生物开始虽然是小说,但仍然有,
176	like Dr.	像医生一样
177	He's phages, a recognizable resemblance to something natural.	他的香肠, 一个可识别的相似的东西自然。
178	But just as natural	但同样自然
179	selection, proceeding one step at a time, went from microbes to humans via fish, amphibians,	选择,一步一步走, 从微生物到人类通过鱼,两栖动物,
180	mammal-like reptiles and monkeys, so AI-driven incremental change might result in applications	类似哺乳动物的爬行动物和猴子,因此AI驱动的增量变化可能导致应用
181	that sound today like fantasy.	今天的声音就像幻想
182	Growing buildings rather than constructing them, perhaps.	而不是建造这些建筑,也许。
183	Delivery	交付
184	drones with wings instead of propellers, off-road cars with legs instead of wheels.	拥有翅膀而非螺旋桨的无人机,有腿而非轮子的越野车.
185	And DNA space itself may prove bigger than it looks at first sight.	而DNA空间本身可能比它乍一看要大。
186	AI-designed artificial cells	AI设计的人工细胞
187	might assist the extension of biochemistry and therefore synthetic biology by rewriting the	可能通过改写生物化学和合成生物学来协助生物化学的扩展。
188	genetic code itself.	基因密码本身。
189	Natural proteins employ only 20 types of amino acid, but DNA would have the	天然蛋白质只使用20种氨基酸,但DNA会具有
190	ability to encode three times as many if the machinery of protein synthesis could be rewired	如果蛋白质合成的机械能被重新连接的话编码能力是三倍
191	to eliminate duplicates that might permit all sorts of useful chemistry that is currently not	消除可能允许目前不存在的各种有用化学的重复
192	seen in nature.	在自然界中被人们看到。
193	It could also, though, unleash accidental monsters.	也可能释放出意外的怪物
194	Dr.	医生
195	He and his colleagues	他和他的同事
196	chose phages for their work on synthetic viruses in part because they pose no conceivable threat	部分因为合成病毒造成无法想象的威胁
197	of turning into pathogens, and the creators of EVO2 made sure that when they were training it,	而EVO2的创造者们确保当他们训练它时,
198	none of the viruses in the databases were pathogenic, lest the model accidentally learn	数据库中没有任何病毒是致病的以免模型不小心得知
199	to do something dangerous.	做一些危险的事情。
200	As the technology spreads, though, risks will proliferate.	然而,随着技术的传播,风险将会增加。
201	Like AI itself, then, the biology that AI will enable offers possible worlds that are both	和AI本身一样,AI能够提供的生物提供了两种可能的世界
202	terrifying.	吓人
203	The next time you read a story about a dragon factory in an issue of The Economist,	下次你读到《经济学家》杂志上有关龙工厂的故事
204	don't be so sure that we're joking.	别那么确定我们在开玩笑
205	Don't be so sure that we're joking.	别那么确定我们在开玩笑

该视频共有字幕 205 条。解锁更多字幕为会员功能，请移动到价格

下载安装