食物链中的人工智能

这个BBC播客由英国境外的广告支持。

This BBC podcast is supported by ads outside The UK.

对食物着迷吗？

Obsessed with food?

喜欢寻找最佳食谱吗？

Love finding the best recipes?

我是塞缪尔·戈德史密斯，美食作家、厨师，也是《Good Food》播客的主持人。

I'm Samuel Goldsmith, food writer, cook, and host of the Good Food podcast.

每周，我都会与顶尖厨师、美食作家以及引领饮食方式变革的人坐下来交谈，这些真正让美食变得出色的人。

Every week, I sit down with top chefs, food writers, and people at the forefront of changing the way we eat, all the people who really make food great.

如果你喜欢

If you love a

一顿美味的餐食和一场精彩的对话，请在Spotify上搜索《Good Food》播客，每周二更新新鲜内容。

good meal and a great conversation, search for the Good Food Podcast on Spotify, serving up fresh episodes every Tuesday.

那里见。

See you there.

从工厂生产线到超市货架，人工智能正在悄然重塑全球食品产业。

From factory floors to supermarket shelves, artificial intelligence is quietly reshaping the global food industry.

从农田到餐桌，整个过程的每一个环节都实现了自动化和人工智能驱动，而我们却对此一无所知。

So from farm to fork, everything, every aspect of the process is automated, AI driven, and we just don't know about it.

它正在优化供应链、减少浪费，甚至帮助厨师在厨房中工作。

It's streamlining supply chains, tackling waste, and even helping chefs in the kitchen.

我大部分时间都会向人工智能寻求单一食材的灵感。

I'm most of the time going to the AI with ideas of one ingredient.

我可以问它：你能帮我找一个好点子吗？

I can ask him, can you help me to find a good idea?

在本期《食品链》节目中，我和鲁梅拉·达斯·古普塔探讨：人工智能如何重新塑造食品产业？

On this episode of the Food Chain with me, Rumela Das Gupta, how is the food industry being re engineered by AI?

我们的机器人是协作型的。

Our robots are collaborative.

如果你触碰它，机器人基本上就会停止运行。

If you touch it, the basically, the robot will stop.

对吧？

Right?

所以它不会伤害你。

So it can't hurt you.

因此我们实际上不需要任何防护装置，机器人和人类可以并肩工作进行生产。

So we actually don't need any guarding, and and, basically, the robot and humans can work side by side to do production.

我们审视这些承诺与陷阱，并听取技术背后的人员分享。

We look at the promises, the pitfalls, and hear from the people behind the tech.

请带我们去龙虾五号桌。

We can go to lobster table five, please.

我在想，把芦笋、酪乳和一些刺山柑很好地搭配在一起。

I'm thinking here about a good combination between asparagus, buttermilk, and maybe some capersleeve.

我考虑用烤甜面包来搭配这道菜。

I was thinking about working this course with some sweet bread roasted sweet bread.

你能帮我找一个适合与这道菜搭配的优质香草吗？

Can you please help me to found an idea of the nice herb that been that can be combined with this?

这是厨师马塔内·扎金。

This is chef Matane Zakin.

他刚刚向一个生成式人工智能聊天机器人寻求帮助，以完善他在法国巴黎米其林星级餐厅Nom创作的一道菜，而他经常这么做。

He's just asked a generative artificial intelligence chatbot for help with a dish he's creating at his Michelin star restaurant, Nom in Paris, France, and this is something he does quite often.

这其实始于四个月前。

It's, it actually started, four months ago.

当时没人那么频繁地谈论AI，但在许多其他行业，大家都在使用它。

Nobody was, so much talking about, AI, but in many other industry, everyone was using it.

我在想，为什么我们不能在厨房里也使用它呢？

And I was thinking, like, why we could not use it also in the kitchen?

我的想法是去测试、尝试，并挑战AI。

And the the idea was to test and to try things and to challenge the AI.

你有像创业日常运营那样的安排，比如每天晚上，所有员工都会把他们的订单发给你。

You have like the daily organization of entrepreneurship that for example, every night all the staff sending you all they order.

所以我可以把所有的订单输入AI，让AI根据供应商类别、克或公斤等分类来整理和汇总订单。

So I can just put all my order on AI and ask the AI to regroup the order and to organize them from category of suppliers, from category of gram, kilogram, stuff like that.

而且它们还能提醒我：警告，厨房里有两个人点了同样的产品。

And also they can just tell me warning, warning, you have two different person in the kitchen that ask you for the same product.

你确定吗

Are you sure that

你真的想要两升这个吗？

you wanna get two time liter of this?

所以我认为我在订货和食材成本上犯的错误更少了。

So I think I make less mistake on ordering and on food cost.

而且主要的话题是关于激发创意。

And also the main subject is about pushing the creativity.

所以我认为，现在在人工智能领域，已经有科学家们多年来在书籍中记录了食物搭配和风味搭配的知识。

So I believe that in EI now, you have about many many years we had scientifics that write down on books about food pairing and taste pairing.

人工智能可以告诉我，樱桃和香豆素搭配得很好，因为它们含有

The EI can tell me that cherry and tonka beans going well together because there is

相同的分子。

the same molecule inside.

例如，我可以看出芦笋和开心果搭配得很好。

I can see for example that asparagus and pistachio are going well together.

我不会让AI为我设计一道菜，但我通常会带着一种食材和另一种食材的想法去询问AI。

I'm not going to ask the AI to create a course for me, but I'm most of the time going to the AI with ideas of one ingredient, a second ingredient.

我可以问他：你能帮我找到一个将这两种食材结合的好点子吗？

I can ask him, can you help me to find a good idea to connect of this ingredient?

他可以给我推荐一种调味料，告诉我应该加一点酸味，从而促使我进一步深化和提升我的主要创意。

So he can give me a condiment, he can tell me you should add a bit of acidity to it, so he can just challenge me a bit more to go higher and higher on my main ideas.

这非常有趣。

And this is quite interesting.

马丁大厨认为，AI是一个强大的工具，因为它能利用庞大的现有知识库。

Chef Martin believes that AI is a powerful tool because it draws on an extensive archive of existing knowledge.

从食谱书、科学研究到数十年的在线食谱，AI都能从中汲取信息。

From cookbooks and scientific research to decades of online recipes, AI can tap into all of it.

对于一位受过传统训练的法国厨师来说，这似乎有些反直觉，但马丁表示他正在拥抱未来。

It might seem counterintuitive for a classically trained French chef, but Martin says he's embracing the future.

我相信大多数厨师都有很强的自我意识，包括我自己。

I believe that most all the chefs have a lot of ego, even me.

但法国厨师是最顶尖的。

But the French chef are the tops.

明白吗？

Okay?

所以这就像是脱离了烹饪传统，然后去这么做。

So it's like going out from the tradition of the cuisine and then doing that.

但我相信我们必须跟上时代，如果整个行业都在使用它，我们就必须使用。

But I believe that we need to live with our time, and if in all of the industry everyone using it, we have to use it.

这能推动我们前进。

And this is like can just push us.

所以目的并不是100%依赖它。

So the idea is not to use it 100%.

目的是把它当作一种能帮助你、推动你，同时让你在创作时感到安心的工具。

The idea is to take it as something that can help you, that can push you, and also can just make you feel comfortable about the creation.

有时候我会用油画作品，然后把油画作品放在Charjipati上，比如Charjipati就会告诉我，嗯，这是个好主意。

Sometime I have the oil picture and I just gonna put the oil picture on Charjipati for example, and Charjipati just gonna tell me, okay, this is a great idea.

调味品能带来酸味。

The condiment will give you acidity.

蔬菜烤制后会带来烟熏风味。

The vegetables, just grilled gonna give smoky flavor.

试着找一款黄油酱。

Try to find a buttery sauce.

我就想，好吧。

I'm just like, okay.

我做对了。

I done it right.

但总有一天，每个人都会用它。

But at one point, everyone gonna use it.

这肯定没错。

This is sure.

需要花多少时间，我完全不知道。

The time that it's gonna take, have no idea.

目前，您正在将人工智能用于实验目的，例如食谱生成、优化或提高效率。

At the moment, you're using AI for, say, in an experimentation capacity, for recipe generation, for optimization, maybe for efficiencies.

但您会考虑将其用于质量控制吗？

But would you consider using it for things like quality control?

您会用它来管理浪费，甚至实现厨房自动化吗？

Would you use it for waste management or even kitchen automation?

因此，我相信在我们的烹饪风格中，在我们所处的高级餐饮和美食学、米其林星级餐厅这一特定行业中，我们永远无法达到那种程度。

So I believe that in our style of cuisine, in in our specific industry of fine dining and gastronomy and Michelin star restaurant, we will never get there.

我相信，这正是我们工作的本质限制——每天亲手完成每一件事，并享受这个过程。

And I believe that this is a limit of the essence of our work to do things every day by hand and to enjoy it.

我不相信机器人能够识别并说，这个产品的质量比那个更好。

I don't believe a robot will identify and say, okay, this product is better quality than this one.

我不相信机器人能闻出一个西瓜，然后说，这个已经熟了可以用了，而那个要放在外面再放二十四小时，因为它还需要一点时间。我认为这就是我们的局限。

I don't believe a robot can smell a melon and say, okay, this one is ready to use, and this one we're gonna leave it twenty four hour outside because it needs a bit more And I believe this is our limits.

但我相信，在大型食品工业中，机器人将取代人类完成许多琐碎的工作，而或许会有更多人投身于我们的高端餐饮行业。

But I also believe that in the big food industry, we're gonna have robots that gonna work for all the small works that have been done by human today, but and maybe we're gonna have more people that gonna come to work in our fine dining industry.

但对于披萨、汉堡之类的食物，我们可能会用上机器人。

But maybe we're gonna have like for pizza and burger and stuff like that, we're gonna have it.

但我们在食物中投入的情感，这是无法通过设置来实现的。

But the emotion, the emotion that we put in things, this is something that you can't it's not a setting that

你

you

投入情感。

put emotion.

而这正是我们每天所做的事情。

And this is what we do every day.

人工智能可以在食品产业链的各个环节中以不同方式应用。

Artificial intelligence can be used in different ways in the food chain.

有些系统是生成式的，能够基于学习到的模式创造新食谱或新创意。

Some systems are generative, creating new recipes or ideas based on learned patterns.

其他则是预测性的，通过分析数据来预测需求、减少浪费或优化生产。

Others are predictive, analysing data to forecast demand, reduce waste or streamline production.

还有具身人工智能，比如一些厨房和工厂中的机器人，能够实时感知、决策和行动。

And then there's embodied AI, like the robots in some kitchens and factories that can see, decide and act in real time.

它们都是人工智能领域的一部分，我们的下一位嘉宾对此非常了解。

They're all part of the AI landscape, something our next guest knows a lot about.

我的名字是塔米娜·DC·魏斯坦。

My name is Tamsin DC Weinstein.

我是开曼群岛人工智能领域的思想领袖。

I'm a thought leader on the AI industry in The Cayman Islands.

我同时也是开曼群岛大学学院的高级总监。

I'm also a senior director at the University College of the Cayman Islands.

我主要从事人工智能及其对全球劳动力和不同行业影响的研究工作。

And I do a lot of work on AI and how it's affecting globally the workforce and different sectors.

目前食品行业最有趣的一点是，它完全由人工智能驱动。

The really interesting thing about the food sector at the moment is it's completely driven by AI.

从农场到餐桌，整个过程的每一个环节都是自动化、由人工智能驱动的，而我们却对此一无所知。

So from farm to fork, everything, every aspect of the process is automated, AI driven, and we just don't know about it.

甚至在最开始的农场，种植的作物、种植方式以及收获方式，都通过人工智能、自动化和机器人技术实现，并贯穿整个供应链，直到我们盘中的食物，我们之所以选择它们，都是因为人工智能算法的引导。

So even in the farms at the very beginning, the produce that is being grown and how it's being grown and how it's being harvested, AI automation robotics, going all the way through the supply chain to the foods that we even get on our plates, we are being driven to choose them because of AI algorithms.

让我举一个消费者方面的例子。

Let me give you a consumer example.

我们所有人都通过手机订购食物。

So we all order foods on our phones.

Uber Eats，所有这些食品配送系统都是如此。

Uber Eats, all of the food delivery systems.

现在，拿起手机打开应用，点一份外卖或叫一份餐送到家里，已经成为我们日常生活的一部分。

It's part of day to day life now that we pick up our phones and we open it up and we order a takeaway or food delivery to our house.

我们以为是自己在做选择，但实际上，人工智能算法在向我们展示该选什么、附近有什么、什么能最快送到我们手中。

We think that we're making that choice, but the AI algorithms are showing us what to choose, what's near us, what's going to get to us quickly.

它会根据它所了解的我们的偏好，为我们推荐选项。

It gives us recommendations based upon what it knows our preferences are.

选择权不再属于我们了。

The choice isn't ours any longer.

实际上是由人工智能驱动的。

It's actually being driven by AI.

这是一个非常有趣但无人意识到的概念：我们正在驱动人工智能，而人工智能也在驱动我们。

And it's a really interesting concept that nobody realizes, that actually we're driving AI, but AI is also driving us.

塔姆森说，人工智能目前正推动着食品行业。

Tamsen says AI is currently driving the food industry.

但这种变化是什么时候发生的呢？

But when did that change happen?

对我来说，新冠疫情是这一转变的巨大推动力。

So COVID for me was a huge driver in this.

如果我们对比疫情前和疫情后的生活，会发现完全不同。

If we look at life before COVID and life after COVID, it's completely different.

为什么？

Why?

因为我们被迫以不同的方式做事。

Because we were forced to do things differently.

因此，原本由人类完成的任务、由人类管理的互动，突然间我们再也无法做到。

So tasks that were once done by humans, interactions that were managed by humans, all of a sudden we couldn't do it.

世界完全停摆，我们必须找到解决方案，以让行业继续运转。

The world completely shut down and we had to find solutions for industry to continue.

而许多解决方案就是机器人和人工智能。

And the solutions in a lot of times were robotics and AI.

因为当人类无法再进行互动时，我们必须找到技术来替代。

Because where humans could no longer interact, we had to find tech to do it.

另一方面，我们对使用技术也变得更加适应。

The other aspect of that is we became a lot more comfortable with using technology.

因此，人工智能最大的障碍是我们自己。

So the biggest barrier to AI is us.

人类才是最大的障碍。

Humans are the biggest barrier.

我们害怕它。

We are scared of it.

我们不相信它能比我们做得更好。

We don't believe that it can do things better than us.

作为人类，我们非常自负，认为我们能比任何其他东西都做得更好。

We are egotistical as a race, and we believe that we can do everything better than anything else.

所以，我们才是最大的障碍。

So we're the biggest barrier.

我们阻碍了这种创新。

We stop that innovation.

而新冠疫情改变了这一点，因为我们不得不依赖它。

And COVID changed that because we had to rely on it.

因此，人工智能在那之后彻底爆发了。

So AI completely exploded after that.

但这是否意味着我们会失去工作？

But does this mean that we're losing jobs?

所以我们并没有失去工作。

So we're not losing jobs.

我们正在增强工作。

What we are doing is augmenting jobs.

这是我的一个主要热情所在。

And this is one of my big passions.

显然，我既在工业界工作，也在教育界工作。

Obviously, I work within industry, but also within education.

如果你看看世界上发生的任何一次工业革命，工作都会发生变化。

If you look at any industrial revolution that's happened in the world, jobs change.

多年前，我们有接线员，他们的工作是连接电话。

So years ago, we had telephonists whose job it was to connect calls.

我们有操作电梯的人。

We had people who operated lifts, elevators.

随着技术和创新的接管，这些工作变得多余，工作也随之变化。

The jobs become redundant because as technology and innovation takes over, the jobs change.

因此，毫无疑问，一些工作会消失，但也会产生新的工作。

So jobs will be lost without a shadow of a doubt, but jobs will be formed.

因此，关键在于增强工作、再培训和提升技能，而不是单纯地失去工作。

So it's about augmenting jobs and retraining and upskilling rather than loss.

人工智能也揭示了长期以来被忽视的问题。

AI is also revealing problems that have long been overlooked.

迪尼·麦格拉思是Zest公司的联合创始人，该公司开发了一项人工智能程序来解决其中一个问题。

Deeni McGrath is the cofounder of Zest, a company that has developed an AI program to address one of them.

当人们想到食物浪费时，通常会想到超市货架上看到的大量浪费，以及打折商品上到处都是的黄色标签。

So when people think about food waste, you often think about the supermarket shelf and how much waste you see, the yellow stickers all over the discounted products.

实际上，制造环节的食物浪费量是零售环节的七倍。

There's actually seven times more food waste at a manufacturing level versus retail.

这数量巨大，但几乎没人对此采取行动，因为这是一种隐蔽的、不吸引人的浪费，不像超市或家庭中的浪费那样直观可见。

It's huge and no one is really doing anything about it because it's this sort of hidden unsexy waste that no one is not as tangible as waste at a supermarket level or waste in the home.

迪尼，你能给我解释一下Zest的人工智能是如何工作的吗？

Can you explain to me, Deni, how Zest's AI works?

以一家饼干制造商为例，最初有不同的输入原料。

So taking a biscuit manufacturer, you've got different inputs to start off with, which are all raw ingredients.

你有巧克力、威化饼干，还有其他各种原料。

So you've got your chocolate, you've your wafer biscuits, and you've got other ingredients.

这些是生产最终产品的输入原料。

Those are the inputs that go into producing the end product.

但要达到最终产品，必须经过一个复杂的生产系统，其中威化会被切割成饼干所需的形状，自然会产生边角料。

But in order to get to the end product, it has to go through a complex production system where the wafer will be cut into the right shape for the biscuit, and you'll obviously have the offcuts.

整个供应链中，这些环节都会通过数据和重量进行追踪，但没有人利用这些信息。

And all of these things are tracked using data and weights throughout the supply chain, but no one does anything with that information.

我们的技术会整合生产线上所有的数据点，监测生产线上的重量、初始原料的投入量与实际产出量的对比。

So what our technology does is it pulls together all of the different data points from the production line, looking at the weight on the line, looking at the volume of raw ingredients to start off with versus the output of what it actually produces.

并利用我们专有的AI算法，尽可能接近实时地监控供应链中的浪费情况。

And it uses our proprietary AI algorithms to monitor this waste in the supply chain in as close to real time as it possibly can.

然后我们可以利用这些数据，分析历史趋势：当你生产X数量时，会有多余多少库存。

We can then use this data and look at historical trends of when you produce X amount, this is how much surplus you had.

因此，你可以开始预测供应链中即将产生的废料数量和价值。

So you can start actually predicting the volumes and values of waste that are going to be coming off the supply chain.

通过这些热点，你可以开始审视供应链中的一些小环节，以提高效率。

And with these hotspots, you can then start looking at potentially changing small things on your supply chain to make them more efficient.

你可以以一种符合食品安全标准的方式进行捕捉。

You can capture it in a way that allows it to be food safe.

目前，这条饼干生产线的90%的废料都被用作动物饲料。

And we can see currently 90% of this biscuit line is going to animal feed.

实际上，很多食物仍然适合人类食用，而我们的算法正是要判断这些食物是否适合人类食用，如果仍然可食用，我们能否将其重新分配回食品系统，比如捐赠给食品慈善机构，或提供给不与他们竞争但可将其作为生产原料的其他食品制造商。

Actually, a lot of the food will be fit for human consumption still, and that is our algorithm's job to work out whether it's fit for human consumption and whether, if it's still edible, we can then reallocate it back into the food system, whether that's to a food charity or back to another food manufacturer that is not competing with them, but could use that as an input for their production process.

训练和运行人工智能系统需要大量的计算能力，这意味着消耗大量能源。

Training and running AI systems takes a huge amount of computing power, and that means energy.

从数据中心到强大的处理流程，所有这些数字基础设施都具有真实的碳足迹。

From data centres to powerful processes, all that digital infrastructure has a real world carbon footprint.

那么，Diiny正在采取什么措施来应对这一问题？

So what is Diiny doing to combat this?

AI本身确实会产生影响，是的。

AI itself does have an impact, yes.

我们正在与谷歌合作开展这项试点项目。

And we are doing this pilot in collaboration with Google.

因此，我们所有的操作都通过谷歌云进行。

And so we're running everything through Google Cloud.

他们一直是我们出色的导师，帮助我们了解如何选择各种方案，以尽可能降低我们的碳排放。

And actually they have been amazing mentors for us to understand what the different options are to keep our emissions as low as possible.

例如，我们使用伦敦电网的电力，因为这是目前英国最可再生的能源。

So for example, we are using energy from the London grid because that's the most renewable energy in The UK at the moment.

我们尽可能采纳他们的建议，确保我们的AI和技术带来的影响尽可能小，以免超过我们在食物再分配方面带来的社会和环境效益。

And we're taking their advice wherever we can to make sure that our impact from an AI and tech perspective is as low as it possibly can be so that it never outweighs the benefit of what we're doing from a social and environmental perspective with the redistribution.

那么，你认为你们正在试点的这个AI工具，最终会像制冷技术一样成为食品行业的标配吗？

So do you think this particular AI tool that you're piloting, it could eventually become standard in the food industry, like say refrigeration did?

我非常希望看到它成为标准。

I would love to see it becoming standard.

实际上，去年有一项法案要求所有大型食品制造商必须报告食物浪费情况，这项法案在欧洲已经通过，但在英国尚未通过。

Actually, was a bill last year to make food waste reporting mandatory for all large food manufacturers, and that's been passed in Europe and it hasn't been passed in The UK yet.

但我们需要达到一个能够监测食物浪费的阶段，以便我们能够

But we do need to get to a place where we are monitoring food waste to enable us to

对食物着迷？

Obsessed with food?

喜欢寻找最佳食谱？

Love finding the best recipes?

我是塞缪尔·戈德史密斯，美食作家、厨师，也是《Good Food》播客的主持人。

I'm Samuel Goldsmith, food writer, cook, and host of the Good Food podcast.

每周，我都会与顶尖厨师、美食作家以及改变我们饮食方式的前沿人物交谈，这些才是真正让食物变得精彩的人。

Every week, I sit down with top chefs, food writers, and people at the forefront of changing the way we eat, All the people who really make food great.

如果你喜欢

If you love a

一顿美味的餐食和一场精彩的对话，请在Spotify上搜索《Good Food Podcast》，每周二更新新鲜剧集。

good meal and a great conversation, search for Good Food Podcast on Spotify, serving up fresh episodes every Tuesday.

到时候见。

See you there.

采取行动应对它。

Take action against it.

你认为Zest AI工具能否通过减少低效问题，从长远来看在更大规模上影响食品定价？

Do you think the Zest AI tool could sort of influence food pricing in the long run at a larger scale by, say, reducing the inefficiencies?

这是个非常好的问题，也是我们非常关注的议题。

It is a very good question and something that we care passionately about.

几年前，麦肯锡做了一项出色的研究，显示食品浪费使食品制造商的生产成本增加了约10%。

McKinsey did an amazing piece of research a couple of years ago, which showed that food waste adds about up to 10% onto food manufacturers' cost of goods.

因此，这确实是一种计入食品生产成本的实际开销。

So that is an actual cost that's factored into making our food.

总的来说，在这种供应链的制造环节中，过剩物资价值超过80亿美元。

In general, there's over $8,000,000,000 worth of surplus in this sort of manufacturing part of the supply chain.

所有这些成本最终都会转嫁到我们这些终端消费者身上。

All of that costs ends up with us, the end consumer.

如果我们能帮助食品制造商将浪费减少50%，就意味着生产成本下降5%，这部分节省最终可能会传递给终端消费者。

So if we can help food manufacturers to reduce it by 50%, you're talking about a 5% change in cost of goods, which could then end up being passed down to the end consumer.

这是我们最终希望实现的目标，也是我们每天前进的动力。

It's an end goal of where we want to get to, but it's something that definitely drives us every day.

您正在收听BBC世界服务的《食品供应链》节目。

You're listening to the Food Chain from the BBC World Service.

我是鲁梅拉·达斯·古普塔，本周我将探讨人工智能如何重塑全球食品供应链。

I'm Rumela Das Gupta, and this week I'm finding out how artificial intelligence is rewiring the global food supply chain.

全球有80亿人每天都需要吃饭。

8,000,000,000 people in the world have to eat every day.

我们必须吃饭，这就是为什么食品行业如此庞大。

We have to eat, and that's why food is such a big sector.

这个行业的产值巨大，从农作物种植、供应链到包装，整个链条中都存在大量低效环节。

It's worth a lot of money, and there's a lot of inefficiencies all throughout the chain, from the growth of crops to the supply chain to the packaging.

塔姆津·迪齐·韦恩斯坦是专注于人工智能在各行业和市场应用的战略顾问，居住在开曼群岛。

Tamzin Dizzy Weinstein is a strategic adviser specializing in the application of AI across sectors and markets and is based in The Cayman Islands.

整个供应链中存在如此多的低效环节，导致每年生产的食品有三分之一被浪费。

There's so many inefficiencies that happen all the way along the chain that a third of the food that's produced gets lost every year.

因此，正如你所见，提升效率的空间巨大，而这正是人工智能正在做的。

So as you can see, the room for efficiencies is huge, that's what AI is doing.

如果我们看一下供应链，人工智能现在可以实现供应链的自动化。

If we look at a supply chain, AI will now automate the supply chain.

因此，机器人正在取代人类管理供应链，使整个过程更加成本高效。

So robots are manning the supply chains rather than humans, making the process a lot more cost efficient.

如果你想想机器人和摄像头如何检测故障或变质的食品，它们比人类要高效得多。

If you think about robotics and cameras looking for malfunctions or looking for foods that are spoiled, it's a lot more efficient than a human doing it.

这对公司来说是成本节约，同时也让食物更快、更便宜、更高质量地摆上我们的餐桌。

So it's a cost saving for the company, and it's putting the food on our plate, quicker, faster, cheaper, and at a much greater level of quality.

那么，配备人工智能的机器人是食品解决方案的未来吗？

So are robotics with artificial intelligence the future for food solutions?

Chef Robotics 是一家总部位于美国西海岸旧金山的初创公司。

Chef Robotics is a startup based in San Francisco on the West Coast Of The United States.

我是Chef Robotics的创始人兼首席执行官拉贾特·帕吉里亚。

I'm Rajat Pagiria, founder and CEO of Chef Robotics.

在Chef，我们制造AI驱动的机器人，帮助食品行业应对劳动力短缺问题，通过部署这些机器人实现生产流程的灵活自动化。

At Chef, we make AI enabled robots that help enable the food industry to kind of overcome the labor shortage, by deploying these robots that essentially help them flexibly automate production.

自动化并不新鲜。

Automation is not new.

它已经存在了几十年。

It's been around for decades now.

AI如何助力这种自动化？

How does AI help with this automation?

是的。

Yeah.

食品行业已经有很多自动化了。

There is a lot of automation in the food industry.

你说得对。

You're right.

它已经存在了四到五十年。

It's existed for forty to fifty years.

问题是它非常僵化。

The thing is it's very rigid.

它非常擅长一遍又一遍地做同样的事情。

It's very good at doing the same thing over and over again.

它不够灵活。

It's not very flexible.

因此，食品行业有一部分本质上属于低品种制造。

So there's a part of the food industry that's essentially like low mix manufacturing.

他们就是这么称呼的。

That's what they call it.

低品种指的是，你有非常高的产量，但只有两三种不同的产品，你可以为每种产品建立一条专用的定制生产线，然后持续运行。

Low mix is kind of like, you know, you have very high volumes and you have two or three different products and you could create a dedicated custom line for each of the products and you run the body along.

想象一下，这就像生产iPhone、Mac、乐事薯片或可口可乐的方式。

Imagine that's kind of how you make like iPhones or Macs or or Lay's chips or Coca Cola.

对吧？

Right?

每个产品都有自己的定制生产线，然后你整天不停地运行它。

You each product has its own custom line and you just run it all day long.

因此，某种分发装置在这种情况下效果相当好。

So some kind of dispenser tends to work pretty well for that.

另一方面，如果你想想像餐食这样的东西，每个人想要的餐食都不同。

On the other hand, if you think about something like a meal, well, everyone wants different kinds of meals.

有素食餐，有无麸质餐，等等等等。

There's vegan meals, there's gluten free meals, etcetera, etcetera.

而每种餐食内部都有四五十种不同的SKU。

And each of those has 40 to 50 different SKUs within that.

所以SKU基本上就是一种新产品。

So an SKU is basically a a new product.

比如一种薯片和另一种薯片的区别。

So like, one kind of chips versus another kind of chips.

对客户来说，这就像是一个新产品。

It's like a new product for the customer.

所以如果你有200种不同的产品，你不可能为每种产品都配备一条专属生产线。

And so if and you have 200 different products, you're not going to have 200 dedicated lines.

你只会拥有大约10条生产线，然后会切换生产线，每条线都具备灵活性。

You're going to have instead like 10 lines, and you're going to change over the line, and each line is going be flexible.

这就是传统自动化失效的地方，你不得不依赖人工，而这正是我们通过Chef能提供帮助的地方。

So that's really where traditional automation doesn't work, where you end up having to use people, and that's really where we can help with Chef.

这项技术的采用情况如何？

How much uptake has this technology had?

如今，我们的机器人遍布美国和加拿大各地。

Today, I mean, we have robots all over US and Canada.

我们今年还计划扩展到英国。

We're actually expanding to The UK this year as well.

我们已经在生产中完成了五千万份餐食，据我们所知，这比其他所有食品机器人公司加起来的总量还要多。

We've made 50,000,000 servings in production, which at least from what we understand is is more than all the other kind of food robotics companies combined.

所以这 definitely 是有效的。

So it's definitely working.

Chef Robotics 为食品行业解决的核心问题是什么？

And what is the core problem that Chef Robotics is solving for the food industry?

我认为有几个层面。

I think there's a few layers.

我们试图解决的最重要的问题可能是劳动力短缺。

Probably the most important problem that we're trying to solve is the labor shortage.

至少在美国，根据劳工统计局2023年我最后一次查看的数据，食品加工领域是美国最严重的劳动力短缺问题。

So at least in The US, it's the number one labor at least in 2023, when I last looked at this data, was the number one labor shortage, according to the Bureau for Labor Statistics, in The US food preparation.

因此，其结果是，我们的客户常常无法达成目标。

So basically the result of this is that our customers are oftentimes not meeting targets.

他们让收入流失，还要求员工加班。

They're leaving revenue on the table, they're asking their staff to work overtime.

所以，这可能是最重要的问题：我们能否部署这些人工智能机器人，它们本质上像人一样灵活，没有人被取代，对吧？

So that's probably the number one problem in terms of which is can we deploy these AI enabled robots that essentially are as flexible as people, nobody's being replaced, right?

这只是额外的，因为人手根本不足以满足产量和需求。

It's just in addition because there's just not enough staff to meet volume and demand.

所以这大概是我们存在的最主要理由。

So that's probably the biggest kind of reason we exist.

还有一些其他原因，我们稍后会谈到，其中之一就是一致性。

And a couple of others that we'll kind of talk about, one is indeed consistency.

一种理解方式是，经营餐饮业务的50%成本当然是原材料。

One way to think about this is that 50% of the costs of running a food business is of course the raw ingredients.

而人类往往会过度投放食材。

And humans tend to essentially always over deposit.

这是整个行业普遍存在的趋势，称为‘浪费’。

That's a general trend across the industry, called giveaway.

人们总是多放食材，因为他们不想给顾客提供太少的食物，以免顾客流失。

People are always over depositing because they don't want to serve their customers less food and maybe the customer churns.

对吧？

Right?

因此，结果是这些食品公司总是因食物浪费而损失大量资金。

And so the result of that is that these food companies are always kind of losing a lot of money on food wastage.

而这一点正是我们能做得很好的地方，因为我们的机器人配备了称重系统。

And that's something that we can really do a good job of because our robots have scales, weigh scales integrated into the robots.

因此，每次机器人抓取时，都会测量所用的重量，并进行调整。

And so every single time the robot picks, it measures how much weight it costs, it can adjust.

第三个我想谈谈的是生产效率。

And the third one I'll kind of talk about is throughput.

生产效率之所以重要，是因为你要记住，我们部署的这些生产间通常是冷藏室。

The reason throughput is notable is like, you know, you have to remember that these production rooms that we're deploying into are usually cold rooms.

它们的温度通常在1到2摄氏度，即34华氏度，而你一整天都在重复这些动作。

They're usually one to two degrees Celsius, 34 degrees Fahrenheit, and you're kind of doing these redundant motions all day long.

顺便说一句，这就是劳动力短缺的原因，对吧？

By the way, that's the reason there's a labor shortage, Right?

所以这是一份非常艰苦的工作。

So it's a really tough job.

由于这份工作非常辛苦，有人早上来的时候可能充满干劲，心想今天一定要好好干。

And because it's a really tough job, somebody might come in the morning and be really excited about work and, hey, I'm gonna do a really good job.

但到了下午五点，他们已经疲惫不堪、无精打采，生产效率也变得起伏不定。

And then now we're five, they're tired, they're lethargic, and the throughput is kind of all over the place.

而对于厨师机器人来说，就像节拍器一样稳定。

Whereas for Chef, it's kind like a metronome.

我们可以持续保持高效的净生产率，真正帮助这些公司提升产量。

We can just kind of maintain a really good net throughput and really help these companies increase production volume.

你能给我举个例子吗？你们和客户合作后，大概帮他们节省了百分之多少的成本？

Can you give me an example of what sort of percentage of money that you've saved for your customers that you've been working with?

比如说，厨师机器人的成本比人工便宜20%。

Let's say Chef is 20% cheaper than people.

这没问题。

That's fine.

这很棒。

That's cool.

但这并不会对他们产生实质性的推动作用。

But it's not going to really push the needle meaningfully for them.

但如果我能说：嘿，你们面临劳动力短缺。

But if I can say, Hey, you you guys have a labor shortage.

你们有十条生产线，但任何时候只有八条能运行，因为你们人手不足。

You have 10 production lines, but only eight of them are able to run at any given moment because you don't have enough staff.

如果我现在能让第九条和第十条生产线运转起来，那就能为他们带来大量的额外收入和产量。

And if I can now run lines nine and ten, I mean, that is a huge amount of more revenue and volume I'm able to make for them.

所以我认为，真正能帮助客户提升产量和收入的企业才是最好的企业。

So I think I think my perspective is the best businesses are the ones that really help increase volume and revenue.

这才是最好的企业模式，而这正是我们为客户提供服务的方式。

Those are the kind of the best businesses, and that's kind of what we do for our customers.

这些机器人能够独立工作吗，还是需要有人在旁边协助？

Are these robots able to work in isolation, or do they need people around?

这些装配线的结构基本上是一条长长的流水线。

The way that these assembly lines are structured is basically you have a long assembly line.

在生产线的起点，可能有人放上托盘、卷饼、卷饼或三明治等，然后后面每个人负责添加一种配料。

At the beginning of the line, might have somebody putting down a tray or a burrito or a wrap, a sandwich, what have you, and then each person after that has an ingredient.

也许有人负责加米饭，有人负责加豆子，等等。

Maybe there's somebody putting on rice and somebody putting on beans, etcetera.

对吧？

Right?

在生产线的末端，有人负责密封托盘，或者包裹卷饼之类的操作。

And at the end of the line, somebody's kind of, sealing the tray or maybe wrapping the burrito or something like this.

对吧？

Right?

此外，通常还有一人负责补料。

And in addition to that, there's usually somebody who's doing refill.

他们负责为人工操作的托盘补充物料。

So they're refilling the human trays.

对吧？

Right?

他们只是说：嘿。

They're just saying, hey.

你的鸡肉丁用完了。

You're out of your diced chicken.

让我加点鸡肉丁。

Let's let me put some more diced chicken.

我们设计Chef的方式就是为了适应这个流程。

So the way we've kind of designed Chef is to kind of fit into this workflow.

Chef的整个目的其实是部分自动化。

The whole point of Chef is actually partial automation.

我们并不是想自动化整个流程。

We're not trying to automate the entire process.

对吧？

Right?

有一些自动化类型，比如你想象一个机器人把所有箱子放到托盘上，每条生产线只有一个这样的机器人，所以它是一个单点故障。

There are some kinds of automation, like if you imagine, like, a robot that, like, puts all the boxes into a pallet, there's only one of those per line, so it's a single point of failure.

如果那个机器人坏了，整条生产线就会停摆。

If that robot fails, the line goes down.

而对于Chef，你可以想象有五个机器人，十个人灵活协作完成组装，我们的机器人是协作型的。

Whereas for Chef, you can imagine having five robots, 10 people flexibly working together to do assembly, and our robots are collaborative.

我的意思是它们具有安全等级认证。

What I mean by that is they're safety rated.

所以如果你碰到它，机器人基本上就会停止运行。

So if you touch it, the basically, the robot will stop.

对吧？

Right?

所以它不会伤害你。

So it can't hurt you.

因此我们根本不需要任何防护装置，机器人和人类可以并肩工作进行生产。

So we actually don't need any guarding, and and, basically, the robot and humans can work side by side to do production.

这正是我们的产品为客户带来的真正优势。

And that's really the the beauty of the product for our customers.

你觉得机器人最终会成为标准的厨房工具吗？

And do you see eventually robotics becoming a standard kitchen tool?

我认为这正是我们的愿景。

I think that is kind of the vision.

所以我们的想法是，好吧。

And so our thinking was, okay.

让我们先从高产量的环境开始，也就是食品制造。

Let's start in the high volume environments, which is food manufacturing.

如果我们能在高产量的制造环境中起步，部署大量机器人，你觉得会怎样？

If we can start in the high volume manufacturing environments, we deploy lots of robots, and guess what?

我们会获得真实的生产训练数据。

We get real production training data.

我们会学会如何处理食物。

We learn how to manipulate food.

对吧？

Right?

我们学会如何在不压碎的情况下捡起一颗蓝莓。

We learn how to pick up a blueberry without crushing it.

我们学会如何在托盘上涂抹奶酪而不让它结块。

We learn how to spread cheese in a tray without it clumping up.

我们学会如何捡起几根洋葱，并将它们放入沙拉的特定隔层中。

We learn how to pick up a few strands of onions and place it into a particular compartment of a salad.

我们学会如何操作食物。

We learn how to manipulate food.

如果我们学会了如何操作食物，现在就可以回到低产量的环境，比如餐厅、快餐店、幽灵厨房、监狱、酒店等等，对吧？也就是更日常的商业厨房，然后说：嘿，看，即使你每天只操作四小时，我们现在也能处理所有食材了。

If we learn how to manipulate food, now we can go back to lower volume environments, whether that's restaurants or fast casuals or ghost kitchens or prisons or hotels or what have you, right, more day to day commercial kitchens, and say, hey, look, now we can already do all ingredients even though you're only doing assembly for four hours.

对吧？

Right?

先从高产量开始，随着时间推移再进入低产量领域。

Start with high volume and over time get to low volume.

你认为厨师机器人有朝一日会取代人类厨师吗？

And do you think Chef Robotics will replace human chefs someday?

我不这么认为。

I don't think so.

我认为这正是关键所在。

And I think that's kind of the point here.

对吧？

Right?

我们作为厨师一直在思考的一件事是：让人类做人类最擅长的事，让机器做机器最擅长的事。

One of the key kind of things we think about as chef is like, let humans do what humans do best and let's have machines do what machines do best.

事实上，当厨师是非常有创造力的，对吧？

The fact of the matter is that like being a chef is extremely creative, right?

他们说：好吧，这里有这么多食材，让我从零开始创作一些东西。

They're saying, okay, here's a bunch of ingredients, let me create something out of scratch.

我觉得这种理念非常独特，我们也希望保留它，对吧？

I think that idea is really special and I think we hope to preserve that, right?

我们希望并相信，这将在未来成为一项非常重要甚至更加重要的工作，因为越来越多的日常任务将被自动化。

And we hope that, we believe that that's going be a really important job in the future and perhaps more important as more of these other day to day tasks become more automated.

事实上，人们不喜欢一整天都切菜。

The reality is that people don't like to cut food all day.

他们也不喜欢一整天都做饭。

They don't like to cook food all day.

他们也不喜欢一整天都组装食物。

They don't like to assemble food all day.

对吧？

Right?

所以，如果我们能部署机器人来做这些工作，那么厨师就有更多空间和时间去从事真正的烹饪工作。

So if we can deploy robots to do those jobs, then hopefully we actually have more space and time to do the actual culinary work for the chef.

但这种创新是否只惠及资金雄厚的大企业？

But does this kind of innovation only benefit big players with big money?

小型生产商是否被甩在了后面？

And are smaller producers getting left behind?

以下是人工智能专家塔姆津·迪齐·魏因斯坦的再次发言。

Here's AI expert Tamzin Dizzy Weinstein again.