Can Artificial Intelligence Be Applied to Logistics?

SupplyChainBrain
10 Aug 202113:07

Summary

TLDR在这段访谈中,美国军事大学在线的物流学教授奥利弗·海奇皮斯(Oliver Hedgepeth)讨论了人工智能(AI)在物流领域的应用。他提到自1985年以来,AI已经无处不在,从餐厅、家庭到军事和民用领域都有其身影。AI帮助士兵执行任务,机器人协助战斗,甚至在外科手术和餐饮业中发挥作用。海奇皮斯教授还提到了AI在物流中的具体应用,如亚马逊使用AI和机器人来优化订单处理和配送。尽管AI在物流中取得了巨大进步,但仍存在信任问题,人们对AI系统的偏见和可靠性持有疑问。他强调,尽管AI能够解决问题,但人类仍然需要对AI的决策进行判断和监督。

Takeaways

  • 🧠 人工智能(AI)已经无处不在,从家庭到军事领域都有广泛应用。
  • 🤖 机器人技术是AI的一个重要分支,军事和民用领域都有其身影。
  • 🛒 亚马逊等公司在物流领域使用AI和机器人技术,提高了效率和准确性。
  • 📈 AI在供应链管理中发挥作用,帮助追踪产品和解决延迟问题。
  • 🔍 AI的定义随时间变化,现在有多种定义,取决于个人的观点和背景。
  • 🛑 信任成为AI应用的一个新限制,人们对AI的信任程度不同。
  • 🧐 人们对AI的偏见和背景会影响AI系统的构建和问题解决方式。
  • 🎵 AI可以快速创造音乐效果,显示出在创造性任务中的潜力。
  • 🤔 随着AI的发展,人们对于是否完全依赖AI的决策存在疑问。
  • 🔑 人类判断在AI决策中仍然扮演重要角色,不能完全交给机器。
  • 🚫 AI系统存在局限性,并非所有任务都能完美执行,如机器人引导逃生的失败案例。

Q & A

  • 人工智能在物流领域的应用是如何开始的?

    -人工智能在物流领域的应用始于1985年左右,从那时起就开始了对其应用的探索,至今已经广泛应用于各个领域。

  • 奥利弗·海吉佩斯教授如何看待人工智能在物流中的作用?

    -奥利弗·海吉佩斯教授认为人工智能在物流中的作用无处不在,它帮助士兵执行任务,也在民用领域,如餐厅、家庭和在线购物等方面发挥着重要作用。

  • 人工智能的定义是如何随时间变化的?

    -人工智能的定义随着时间的推移而发生了变化。最初,它旨在模仿人类大脑的思维方式,而现在它更多地被定义为日常使用的实用性技术。

  • 机器人技术在人工智能中扮演什么角色?

    -机器人技术是人工智能的一部分,它们在军事和民用领域都发挥着重要作用,如帮助战斗、侦察、携带物资等。

  • 在物流领域,人工智能有哪些令人兴奋的应用?

    -在物流领域,人工智能用于跟踪库存、优化供应链、自动化仓库管理、快速处理在线订单等,极大地提高了效率和准确性。

  • 人工智能在物流应用中的限制是什么?

    -人工智能在物流应用中的限制包括信任问题、偏见问题以及人类对完全自动化决策的接受程度。

  • 为什么信任在人工智能应用中如此重要?

    -信任对于人工智能至关重要,因为人们需要相信AI系统提供的解决方案是正确和可靠的,尤其是在关键决策和行动中。

  • 人工智能系统是否存在偏见,如果是,它们是如何产生的?

    -是的,人工智能系统可能存在偏见,这些偏见通常来源于创建系统的人的背景、经验和价值观,这可能导致AI系统在解决问题时产生不同的结果。

  • 人类在人工智能决策中扮演什么角色?

    -人类在人工智能决策中扮演着关键角色,他们需要对AI系统的输出进行判断和选择,以决定是否执行AI的建议。

  • 人工智能在预测和规范性建议方面的未来发展方向是什么?

    -人工智能正从描述性系统发展到预测性系统,最终可能发展为规范性系统,这意味着AI将能够提供具体的行动建议。

  • 我们是否应该完全信任人工智能系统,并且不关心它是如何得出结论的?

    -尽管人工智能系统在某些方面比人类更高效,但我们仍需要人类进行最终的判断和决策,以确保AI系统的建议是安全和合适的。

  • 人工智能在紧急情况下的应用有哪些局限性和风险?

    -在紧急情况下,如火灾逃生,人工智能可能无法完全可靠地指导人们安全撤离,这要求人们不能完全依赖AI,而需要保持警惕和自主判断。

Outlines

00:00

😀 人工智能在物流中的应用

本段落介绍了人工智能(AI)在物流领域中的应用。奥利弗·海吉佩斯教授,作为美国军事大学在线物流学教授及陆军人工智能物流中心的创始主任,分享了自1985年以来他在AI物流应用方面的经验。他指出AI无处不在,无论是在餐厅、家庭、军事还是民用领域都有广泛应用。AI帮助士兵执行任务,机器人和AI在医疗、餐饮等领域的应用也越来越广泛。此外,还讨论了AI的定义随时间的演变,以及人们对AI的接受程度和信任问题。

05:01

🤖 机器人技术与AI的结合

这一部分深入探讨了机器人技术在军事和民用物流领域的应用。军事领域中,机器人帮助作战,使用无人机侦察和打击敌人,以及在夜间帮助士兵避开地雷。民用领域中,AI和机器人在物流系统中扮演着重要角色,如亚马逊使用AI和机器人挑选和打包货物,以及在仓库管理中跟踪产品。AI还帮助处理由于疫情导致的供应链延迟问题。然而,AI在物流领域的应用也存在局限性,主要是人们对于AI和机器人的信任问题。

10:01

🔍 AI在物流中的局限性和人类的角色

在这一部分中,讨论了AI在物流中应用的当前局限性,特别是信任问题。美国国家标准技术研究所(NIST)的AI小组研究了人们对机器人和AI系统的信任,并探讨了AI的不同定义和类别。AI系统可能因为开发者的背景和偏见而不同,这影响了人们对AI的信任。尽管AI在解决问题方面取得了进展,但人类判断在执行AI系统的结论时仍然扮演着重要角色。此外,AI系统的决策过程是否透明,以及人们是否接受由“黑箱”直接输出的结论,也是值得关注的问题。

Mindmap

Keywords

💡人工智能

人工智能(AI)是指由人造系统所表现出来的智能行为,包括学习、推理、自我修正、知识理解、语言识别等。在视频中,人工智能被广泛应用于物流、军事、家庭、医疗等领域,如帮助士兵执行任务、在餐厅和家庭中使用的智能机器人等。

💡物流

物流是指商品从生产地到消费地的整个流动过程,包括运输、储存、装卸、搬运、包装、流通加工、配送、信息处理等环节。视频中提到,人工智能在物流领域的应用非常广泛,如亚马逊使用AI和机器人进行商品分拣、跟踪订单等。

💡军事应用

军事应用是指将人工智能技术应用于军事领域,以提高作战效率和安全性。视频中提到,军事领域使用AI来帮助士兵执行任务,如使用机器人进行战斗、侦察和排雷等。

💡机器人

机器人是一种能够执行人工任务的机器,它可以是自动的或半自动的,并能在没有人直接操作的情况下运行。视频中提到,机器人在军事和民用领域都有广泛应用,如帮助士兵在战场上侦察敌情,或在家庭中作为清洁助手。

💡信任

信任在视频中被提及为人工智能应用中的一个重要因素。随着AI的普及,人们对机器人和AI系统的信赖程度变得至关重要。例如,人们是否信任AI系统给出的建议或执行的任务,这影响了AI技术的接受度和应用范围。

💡偏见

偏见指的是由于个人背景、文化、经验等因素而形成的对某些事物或人群的倾向性看法。视频中提到,人工智能系统可能会因为开发者的偏见而产生不同的输出结果,这影响了人们对AI系统的信任。

💡供应链

供应链是指从原材料获取到最终产品交付给消费者的整个流程。视频中提到,AI在供应链管理中扮演着重要角色,尤其是在疫情期间,供应链面临延迟,AI帮助追踪产品流动和预测可能的延误。

💡预测系统

预测系统是一种能够基于现有数据预测未来事件或趋势的AI系统。视频中提到,AI正从描述性系统发展到预测性系统,这意味着AI不仅能分析现有数据,还能预测未来可能发生的情况。

💡规范系统

规范系统是指能够提供具体行动建议的AI系统。视频中讨论了AI系统从描述性、预测性发展到规范性的过程,规范性系统能够告诉人们应该采取哪些具体行动来解决问题。

💡黑箱

黑箱是指那些其内部工作机制不透明或难以理解的系统。视频中提到,随着AI系统变得更加复杂,人们可能会对这些“黑箱”系统如何得出结论感到困惑,这影响了人们对AI的信任和接受度。

Highlights

人工智能(AI)在物流领域的应用非常广泛,包括军事和民用领域。

AI在日常生活中无处不在,如餐厅、家庭、军事和民用领域。

AI帮助士兵执行任务,如机器人和无人机的使用。

AI在民用领域也有广泛应用,如亚马逊物流系统。

AI在物流中用于跟踪产品,应对供应链延迟问题。

AI的定义随时间变化,不同人有不同的理解。

机器人技术在军事和民用领域同样无处不在。

AI在物流中用于解决复杂问题,如供应链跟踪。

AI的局限性之一是人们对其的信任问题。

美国国家标准技术研究院(NIST)研究了人们对AI的信任问题。

AI的偏见问题会影响人们对其的信任。

人类判断在AI系统得出结论后仍然很重要。

AI系统的发展从描述性到预测性再到规范性。

人们对AI系统的完全信任可能导致问题。

AI系统在特定领域解决问题的能力有限。

人类在AI系统发展中仍扮演着重要角色。

Transcripts

play00:00

[Music]

play00:08

how is artificial intelligence being

play00:10

applied to logistics today my guest is

play00:12

oliver hedgepeth he is a pro professor

play00:15

of logistics at the american military

play00:18

university online hello oliver how are

play00:21

you

play00:22

robert i'm doing very well thank you

play00:24

very much for the invitation to talk

play00:26

about ai and it is a big deal here and

play00:29

you are certainly an expert as founding

play00:31

director of the army's artificial

play00:33

intelligence center for logistics how do

play00:35

you see the role of ai in logistics

play00:37

today

play00:38

well

play00:39

started with that job about 1985

play00:42

and been working

play00:45

applications of it for ever it seems

play00:47

like but ai is everywhere today you

play00:50

don't see it because it's everywhere

play00:52

we do have it at restaurants we have it

play00:55

in your home you've got vacuum cleaners

play00:58

uh in the military and in the civilian

play01:00

world it's rampant

play01:01

there are

play01:03

people people don't realize how much the

play01:05

military is using ai artificial

play01:07

intelligence to help soldiers do their

play01:09

job

play01:10

and robots as well and surgeons are

play01:12

using it and restaurateurs are using it

play01:15

cooks are using it sometimes so it is

play01:18

everywhere it's expanding

play01:20

and it's going to continue to expand

play01:22

it's here with us and there are people

play01:24

who don't like it or don't like a robot

play01:27

or don't like smart systems saying

play01:29

here's what you should do

play01:31

such as when you type

play01:32

in a word document on your computer and

play01:35

you misspell a word and it changes it

play01:37

for you because it knows you've

play01:39

misspelled it that's artificial

play01:41

intelligence people don't realize that

play01:43

so that's that's interesting

play01:44

yeah that's such a wide definition

play01:47

i think maybe the definition of ai has

play01:49

changed over the years back decades ago

play01:51

when it was first being developed the

play01:53

idea was to somehow replicate the way

play01:55

that the human brain

play01:57

thought and we were always like 10 years

play01:59

away from ai becoming a reality and now

play02:01

all of a sudden it's everywhere but

play02:03

maybe under slightly

play02:04

different definition that it's more like

play02:06

everyday use usefulness now right

play02:09

well yes it is like everyday use and

play02:12

you're also correct on the definitions

play02:14

changed i've got a nice computer

play02:16

dictionary published in 1967 when i

play02:19

started in the computer world and it

play02:22

defined artificial intelligence it

play02:24

seemed to be pretty good definition

play02:27

but today it has changed and i did look

play02:29

up the definition of artificial

play02:31

intelligence

play02:32

um a few days ago for a paper i'm

play02:35

writing and i found six different

play02:37

definitions it's like really that's kind

play02:40

of cool because the definitions depend

play02:42

on who you are robert you have a

play02:45

different view of what ai might mean to

play02:46

you i have one for me we're biased in

play02:49

how we do this there is no one

play02:52

definition

play02:53

of ai merriam-webster the dictionaries

play02:55

do have a definition but there really

play02:57

isn't just one definition there's

play02:59

varieties of it and it changes

play03:02

almost every year now yeah well i'm

play03:04

interested in what it means to you so

play03:05

that's kind of the

play03:07

subject of this conversation today now i

play03:09

know that ai and robotics are not

play03:11

completely synonymous but where does

play03:13

robotics come into the picture when

play03:14

we're talking about ai

play03:17

a lot of people don't realize that

play03:19

robotics is everywhere too and

play03:22

the military as i mentioned earlier the

play03:23

military and civilian world the military

play03:26

for example has

play03:28

robots that are helping uh fight battles

play03:30

for them they help them shoot the enemy

play03:32

we have robotic

play03:34

drones that can go out and spot an enemy

play03:37

and shoot them send a missile to them we

play03:40

have robots in the military field that

play03:42

when a soldier is walking through high

play03:44

grass field with his weapon at night and

play03:46

night cargo is trying to look for the

play03:48

enemy

play03:49

well guess what there's a little robot

play03:51

out there with him

play03:52

doing the job for him just a little

play03:54

ahead of him to make sure the man

play03:58

doesn't walk on a mine or the woman

play04:00

soldier doesn't walk on a mine and the

play04:03

robot does it we have robots that will

play04:05

carry the soldiers

play04:07

effects you know instead of having to

play04:08

load all these hundred pound bags on

play04:10

your back well there are robots to help

play04:12

do some of that as well so robots are

play04:15

everywhere yeah

play04:16

okay civilian world too yeah well let's

play04:19

talk about the civilian world what are

play04:20

some of the specific uses in in

play04:22

logistics

play04:23

specifically some really exciting and

play04:26

really far-reaching applications of ai

play04:28

in that area right now oh gosh that that

play04:30

is a wonderful field

play04:32

people think of logistics as

play04:34

well

play04:35

ordering something from amazon okay

play04:38

logistics and amazon works anonymous

play04:42

amazon uses ai and uses robots to pick

play04:47

boxes up that they're going to put in a

play04:49

box and send to you along with the human

play04:51

that's next to them they use a lot of

play04:53

robots they use thousands of robots and

play04:56

it's part of the logistics system they

play04:58

use ai to track who you are

play05:00

if you go order something online

play05:03

your and you pay it by credit card the

play05:06

software is ai based is instantly

play05:08

checking everything and making sure you

play05:10

get what you want and the delivery

play05:12

schedule comes to you within seconds

play05:14

where that didn't happen before that's

play05:16

ai that's doing that

play05:17

so warehousing is using a lot of it

play05:21

you go to a grocery store they use ai to

play05:24

help track

play05:25

the products that are there in the back

play05:27

that are on a supply chart coming from

play05:29

some distance far away or they may be

play05:31

delayed now that we are living in a

play05:33

pandemic world this is 2021

play05:36

there are six months to 12 months delays

play05:39

on products coming to retail stores to

play05:42

restaurants

play05:43

to grocery stores to clothing stores and

play05:46

so there's ai that's tracking where the

play05:49

supply chain is because the human has

play05:52

too much going on there's the supply

play05:54

chain and the logistics systems are

play05:56

really too large for one person to have

play05:58

in their head and to keep sorted out

play06:00

you've got too much data for too many

play06:02

thousands there are thousands of supply

play06:04

chains which have parts that go all the

play06:07

way back to maybe a little copper mine

play06:10

in some foreign country that build

play06:13

digs that copper up that goes and

play06:15

finally makes it into

play06:16

that iphone you got yeah and some kind

play06:19

of oil id oil compound that comes from

play06:22

maybe alaska oil fields or

play06:25

iranian oil fields wherever it comes

play06:26

from or texas

play06:28

and it's all linked together it's so

play06:30

it is so complex but it's everywhere and

play06:34

it's most exciting to see it happen well

play06:36

it's come it's come a long way in a very

play06:39

relatively short time but i'm wondering

play06:41

at this point and of course it's still

play06:42

evolving

play06:43

what are the limitations of ai today as

play06:45

applied to logistics wow the limitations

play06:49

today uh

play06:51

are there's a new kind of limitations

play06:53

happening and it's called i'd say trust

play06:55

the national institute of standards and

play06:58

technology it's a department of commerce

play07:00

company okay you don't think of them as

play07:03

ai a lot well they just published a

play07:05

report in june they've got an ai group

play07:08

that's looking at how people trust

play07:10

robots how people trust

play07:12

ai that's helping them do their job live

play07:15

in their home you know vacuum their

play07:17

carpet or something or order something

play07:20

uh a product

play07:22

they have

play07:23

studied and started thinking about why

play07:26

you trust a robot

play07:28

or why you trust ai system

play07:30

and they've defined

play07:32

different aspects of ai you more about

play07:34

talk about definitions they've got like

play07:36

30 different categories of

play07:39

ai

play07:41

definitions and it's like really strange

play07:43

to do that but it's trust and it's in it

play07:46

they've been looking at the trust in

play07:47

terms of bias see you were biased and i

play07:51

am biased based on our background if i

play07:53

build an ai system if you build an ai

play07:56

system and i build one doing the same

play07:58

thing

play07:59

they will not be identical because you

play08:01

have a different age a different

play08:03

background you come from california i

play08:05

came from virginia that may make me do

play08:08

something with the computer programming

play08:09

code just a little bit different so that

play08:12

when it's trying to solve a problem

play08:13

because ai is solving problems robots

play08:16

are solving problems it may solve it

play08:18

just a little bit different for someone

play08:20

who's using it and that gets toward the

play08:23

trust factor you may not trust it if

play08:25

it's giving you an answer that's

play08:27

slightly different than you think it

play08:29

should be well

play08:30

you don't really want to trust it

play08:32

entirely do you i mean you don't want to

play08:33

distrust an ai system but isn't there

play08:35

still a role for humans in terms of what

play08:38

they do with the conclusion of an ai

play08:41

system as to how they choose to execute

play08:43

it or not

play08:45

so human judgment

play08:46

yeah

play08:47

human judgment is still there but i was

play08:49

looking at a program uh two or three

play08:52

weeks ago on television of these

play08:54

musicians i can't remember their name

play08:56

and they were all excited they had they

play08:59

put together music and they have all

play09:00

these instruments and they take weeks

play09:02

and weeks and weeks to get a new sound

play09:04

well they bought some kind of ai gadget

play09:07

that makes sound

play09:08

and it takes in a matter of seconds to

play09:10

create this brand new sound effect

play09:13

a new instrument effect that used to

play09:16

take two or three weeks and they say ai

play09:19

is going to solve my problem

play09:20

they trust it 100 percent so there are

play09:23

people like yeah i'm turning it over to

play09:25

you you give me the solution ai machine

play09:28

you give me the music sound

play09:31

i want to hear they tell the ai system

play09:33

here's what i want and they sit back and

play09:35

they trust it 100 percent now not

play09:37

everybody does but there are people who

play09:39

do trust ai 100

play09:42

but as we go as ai evolves from being a

play09:46

descriptive system

play09:47

to a predictive system and finally to a

play09:50

prescriptive system where it's actually

play09:52

telling you what actions to take

play09:55

do we want to leave that up to the

play09:56

machine and if we do want to leave it up

play09:58

the machine do we want to know how it

play09:59

arrived at that conclusion or are we

play10:01

comfortable with this idea of the black

play10:02

box just outputting a conclusion that we

play10:05

should go ahead and implement because we

play10:06

know they're smarter than we are

play10:09

i i'm glad you asked that question uh

play10:11

and that was part of the uh report from

play10:13

the national institute of standards of

play10:15

technology published just last june

play10:18

from their ai task force in which

play10:21

you don't have to worry about that um

play10:24

there was a comment made by some one of

play10:26

the authors about it may be 20 or 30 or

play10:29

40 50 years from now before you get to

play10:32

that stage the ai systems are still

play10:35

they're strictly prescriptive they go

play10:37

from here to here you give them a

play10:39

problem to solve and they solve that

play10:41

problem you know they may be able to

play10:43

play chess with you and beat the heck

play10:45

out of you

play10:46

but they can't clean the carpet

play10:49

okay or they can't recommend uh an

play10:52

ingredient for

play10:53

spaghetti and meatballs they might they

play10:55

can't you know they may not be able to

play10:56

do that they can't do everything it's

play10:58

not like a human brain in the ai system

play11:02

or even a robot that does something

play11:04

they're limited

play11:05

they're very prescriptive it's almost

play11:07

like medicine they're very prescriptive

play11:09

to solve a problem you got a headache

play11:11

you take an aspirin you got a problem to

play11:13

solve that's really unique for this

play11:15

robot or this ai system you turn it on

play11:18

and it gives you a solution

play11:20

but

play11:21

you as the human have to sit back and

play11:23

say

play11:24

do i really want to do that

play11:27

uh i'll give you one example

play11:29

there was a robot that was

play11:31

being used to guide people out of

play11:33

burning buildings it's dark and smoky

play11:37

and people not firemen

play11:39

but people are in there and it's smoky

play11:41

and it's like a big building in downtown

play11:44

new york or somewhere would be on fire

play11:46

and you want a robot to guide you

play11:47

through this dark

play11:49

way to get out the robot was guiding the

play11:51

people into

play11:53

closed rooms

play11:55

you know it didn't work all the time

play11:57

ninety percent of the time it got them

play11:58

out the door you know it got into it

play12:02

like this is not a situation where ten

play12:04

percent is an acceptable margin they're

play12:06

like

play12:07

they got it they read the robot read

play12:09

something wrong and whatever or read

play12:12

something differently maybe i won't say

play12:14

wrong read it differently and it went

play12:15

into a closet with a bunch of people now

play12:18

here's the key thing

play12:20

those people who got into the closet

play12:23

they stayed there

play12:24

thinking well the robot's gonna fix it

play12:26

now i'm gonna wait here he's gonna tell

play12:29

us to go out and go to the right place

play12:30

any minute now we won't go looking

play12:34

wow

play12:35

okay

play12:36

well the message here is that ai has

play12:38

come a long way it does great things but

play12:40

still a role for humans to play as ai

play12:43

even as ai develops uh dr oliver

play12:45

hedgepeth uh i really appreciate whether

play12:48

the american military university online

play12:50

i really appreciate your participation

play12:52

in this interview and your insights into

play12:54

this into this issue thank you so much

play12:56

for being with me today well thank you

play12:58

very much for the questions

play13:06

you

Rate This

5.0 / 5 (0 votes)

Étiquettes Connexes
人工智能物流应用军事AI民用机器人亚马逊物流供应链管理信任问题人机交互技术进步专家访谈未来趋势
Besoin d'un résumé en anglais ?