0:00

Brought to you by Purdue

0:02

University. The only university

0:04

on fast companies brands that

0:06

matter list. Four years running.

0:09

Honored as a heritage brand,

0:11

Purdue pushes innovation and partnership

0:14

to make a consequential impact

0:16

on the world. Learn more

0:18

at Purdue.edu/partner with

0:20

us. I

0:23

think any worst-case scenario

0:26

starts with leaders

0:28

in an organization

0:30

outsourcing the implications of

0:33

AI to AI. Welcome to

0:35

Fast Companies, the new way

0:38

we work, where we take

0:40

listeners on a journey

0:42

through the changing landscape

0:44

of our work lives

0:46

and explain exactly what

0:48

we need to build

0:50

the future we want.

0:52

I'm Fast Company Deputy

0:54

Editor Kathleen Davis. I've

0:56

been covering workplace culture

0:59

for over a decade and

1:01

there are a few topics

1:03

that consistently bring out strong

1:05

feelings. Performance reviews are at

1:07

the top of that list.

1:09

While most people would agree that it's

1:12

a good thing to have a tool

1:14

to measure how employees are doing at

1:16

their jobs and a time for managers

1:18

to discuss career advancement, very few

1:20

seem to think that the

1:22

way performance reviews are currently

1:24

set up is working. In fact,

1:27

a Gallup survey last year found

1:29

that only 2% of human resource

1:31

officers at major companies think their

1:33

performance management system is working, and

1:36

66% of employees feel the same.

1:38

One of the biggest complaints

1:40

employees have about performance reviews

1:42

is that they're so subjective.

1:44

What it takes to be considered good

1:46

at your job or eligible for a

1:49

raise or promotion is often down to

1:51

the opinion of just a couple of

1:53

people, which means it's fertile ground for

1:55

bias. So if both employees and leadership think

1:57

performance reviews are broken... Could artificial

1:59

intelligence be the magic bullet that

2:01

fixes it? The new crop of

2:03

startup selling AI-powered performance management tools

2:05

certainly think so. But is it

2:07

just swapping out human bias for

2:09

AI bias? Is something as nebulous

2:11

as being good at your job

2:13

quantifiable? And are humans ready to

2:15

be evaluated by a robot? To

2:17

help me parse out the benefits

2:19

and drawbacks of using technology in

2:21

both performance reviews and layoffs is

2:23

Brian Ackerman. The head of AI

2:25

strategy and transformation at Corn Ferry,

2:27

a management consulting firm. Brian, thank

2:30

you so much for being here.

2:32

Thanks for having me. So I

2:34

opened this episode with some stats

2:36

on how disliked performance reviews are.

2:38

That's what we're here to talk

2:40

about today. They're both disliked by

2:42

leadership and employees. And one of

2:44

the biggest complaints that we've seen

2:46

in our coverage is that they

2:48

often feel meaningless or only punitive

2:50

and that they're incredibly subjective and

2:52

open to bias. For those reasons,

2:54

it's easy to see why using

2:56

an AI tool could be tempting.

2:58

What's your view on how AI

3:00

can help solve those problems? Well,

3:02

I would not argue with any

3:04

of those findings or that analysis.

3:06

It feels a little bit like

3:08

we've lost the plot on performance

3:10

reviews. You know, who does this

3:12

help really? So I think that

3:14

needs to be the foundation of

3:16

the discussion for where best to

3:18

use AI. Let's pull back for

3:20

a second and say, what is

3:22

the fundamental thing we're trying to

3:24

change? What is a valid, meaningful,

3:26

meaningful input? for the enterprise and

3:28

then quite frankly can we reintroduce

3:30

so to speak a what's in

3:32

it for the employee getting this

3:34

review because those connections you know

3:36

I think I would agree with

3:38

you are pretty tenuous so I

3:40

think that is the entry point

3:42

for the discussion about AI we

3:44

tend to think about this topic

3:46

with a couple of dimensions one

3:48

would be just the official of

3:50

the process, we can talk through

3:52

that, and that's got an employee

3:54

side of that as well as

3:56

a reviewer or a waiter perspective

3:58

and HR perspective. Also, I think

4:00

there's an opportunity to change using

4:02

AI the way performance reviews are

4:04

consumed by the person, to make

4:06

them more understandable, more impactful, more

4:09

relevant, and then probably even more

4:11

importantly turn these back into. development

4:13

tools, turn these back into tools

4:15

that help you as a professional,

4:17

move along in your career, all

4:19

of which is kind of left

4:21

unsaid with the, hey, you're a

4:23

three out of five or a

4:25

one out of two, or you

4:27

know, whatever your organization's approach to

4:29

this happens to be, in box

4:31

seven of a nine box. Yeah,

4:33

I mean, it's exactly that. I

4:35

think one of the things that

4:37

a lot of people hate about

4:39

performance reviews is that they do

4:41

seem meaningless in that like, what

4:43

the hell does a three mean

4:45

versus a two? What does like

4:47

super exceeds expectations versus exceeds expectations

4:49

mean? I guess on like a

4:51

basic level, when you think of

4:53

AI, it's data driven. Is there

4:55

a pitfall or chance that AI

4:57

kind of goes too far in

4:59

that direction and turns everything in

5:01

what feels like a very human

5:03

process, right into just a set

5:05

of data? Absolutely, exclamation point. You

5:07

can kind of look at these

5:09

three categories in turn. From an

5:11

efficiency perspective, let's talk about that

5:13

for a minute. Tools now, we'll

5:15

use General AI to help draft,

5:17

have the manager draft notes, which

5:19

is all, okay, subject to the

5:21

same challenge of what the quality

5:23

of the draft is and, you

5:25

know, how much work I'm going

5:27

to have to do to make

5:29

that actually meaningful as a manager.

5:31

Am I saving any time anyway?

5:33

Right. allow a manager to have

5:35

a bigger, more robust set of

5:37

data, maybe engagement data, maybe operational

5:39

data to add to the mix

5:41

when they're crafting a report. Another

5:43

good use case, but as you

5:46

said, absolutely subject to the quality

5:48

and accessibility of that kind of

5:50

data that can go and create

5:52

more of a problem. than it

5:54

solves if the data quality isn't

5:56

great. The question of how are

5:58

we making a manager's life easier

6:00

or harder, and since a part

6:02

of the challenge here is with

6:04

respect to managers of the world,

6:06

the ones who kind of do

6:08

it by road. right, who aren't

6:10

really investing in the time to

6:12

form a basis for a good

6:14

performance review, is this going to

6:16

help or hurt? All of that

6:18

around is generative AI and adding

6:20

more data into the mix, you

6:22

know, making this process easier and

6:24

more efficient and more consistent and

6:26

safer and less bias? Or is

6:28

it just adding complexity that then

6:30

the manager is somehow supposed to

6:32

still make sense around? Yeah, when

6:34

we talk about kind of adding

6:36

complexity and the manager role in

6:38

it, I'm a manager. I use

6:40

a software system for... our performance

6:42

reviews and of course like everything

6:44

else it's popped up when I'm

6:46

writing something and I'll say like

6:48

make this better with AI and

6:50

as a writer I never use

6:52

it of course you know I'm

6:54

like no you will not make

6:56

me better with AI but I

6:58

wonder from an employee standpoint does

7:00

it take subjectivity and bias I

7:02

really want to dig into and

7:04

that is certainly a complaint with

7:06

performance reviews but does it feel

7:08

differently as an employee than to

7:10

get a performance review that's written

7:12

by a bot and maybe better

7:14

in a good way, maybe because

7:16

it takes that subject to video.

7:18

I think we have to be

7:20

cognizant that it's February 12, 2025,

7:23

and this technology is as bad

7:25

as it's ever going to be,

7:27

right? It only gets better from

7:29

here. So right this second, if

7:31

you go into those tools and

7:33

they're drafting language, they will come

7:35

across a bit stilted, a bit

7:37

sterile, right? In the interest of

7:39

pulling... bias out of it or

7:41

at least attempting to, of being

7:43

consistent, they're absolutely coming across as

7:45

sterile and need a lot of

7:47

input. Does that particular element change

7:49

as these tools improve as you

7:51

start seeing the new models make

7:53

their way into these products? probably,

7:55

right? So you could see that

7:57

the quality of these drafts are

7:59

going to improve, but I think

8:01

that is different from that it's

8:03

going to not sound or act

8:05

like it's written by a bot.

8:07

We are skeptical that this is

8:09

going to be good enough that

8:11

you're still not going to need

8:13

essentially the same human review that

8:15

you want, especially as you said.

8:17

this process is screaming for better

8:19

human interaction. So I think it'll

8:21

improve, but I'm not sure that

8:23

it's really going to become any

8:25

significantly easier, as it will in

8:27

some other areas where you're using

8:29

generative AI to generate content. Yes,

8:31

it sounds like you're saying, in

8:33

which makes a lot of sense.

8:35

Is this a performance review? Is

8:37

it? essentially human process. AI in

8:39

the best case scenario can help

8:41

improve it, but it's not going

8:43

to replace it. You do not

8:45

envision a future where my performance

8:47

review is with a computer and

8:49

not with a human at all.

8:51

Yeah, that was easy, at least

8:53

for me. No, I hope not,

8:55

and I kind of dread the

8:57

day that we go down that

8:59

path. That is not to say

9:02

that you can't give the human

9:04

superpower, you can't... give them access

9:06

to a broader swath of data

9:08

to make them a better manager

9:10

and do a better review, but

9:12

it's still the person providing feedback

9:14

to another individual. As I said,

9:16

I think you got to keep

9:18

coming back to why are we

9:20

doing this process again? We've lost

9:22

the plot. If the purpose is,

9:24

as it should be, to provide

9:26

meaningful guidance that addresses the individual's

9:28

growth and the organization's requirements in

9:30

the context of the performance of

9:32

their role, then let's go do

9:34

that. And none of that screams,

9:36

let me take the human out

9:38

of the equation and give it

9:40

all to a generative AI capability.

9:42

You mentioned that today in 2025,

9:44

these tools are not very advanced,

9:46

but they obviously like all of

9:48

these. get more advanced. Are there

9:50

specific tools that you've seen that

9:52

are more common than others, things

9:54

that people are using, and kind

9:56

of what are the challenges and

9:58

opportunities with those? Every performance management

10:00

module of every HR tech software

10:02

company is introducing, just as you

10:04

said, Kate, the hey, make this

10:06

better with AI, that's really becoming

10:08

commonplace very quickly. There is a

10:10

generation of bespoke tools coming up

10:12

from employee engagement world that are

10:14

using generative AI to aggregate information

10:16

together for the reviewer to make,

10:18

you know, surface insights from data

10:20

as part of that drafting process.

10:22

To be honest, I think we

10:24

see the most. use in the

10:26

space in the general purpose tool,

10:28

so the Microsoft copilots, the chat

10:30

GPTs, right, the ones that were

10:32

also using to create presentations and

10:34

word documents and emails these days,

10:36

that's still the most common use,

10:39

and then cutting and pasting it

10:41

into reviews. But they're all that

10:43

GPT-40 generation of underlying foundation model,

10:45

so they all act that way.

10:47

And that's the part as the

10:49

models get reintegrated into the HR

10:51

text act. the quality of the

10:53

answer will improve, but right now

10:55

it's a drafting tool. You could

10:57

even argue, I know not all

10:59

companies do this, but a lot

11:01

of companies then do legal and

11:03

compliance reviews on this stuff. There's

11:05

no indication that that's going to

11:07

go away, right, because the level

11:09

of quality of the results isn't

11:11

quite there to pass muster on

11:13

a legal review either. I can

11:15

see a use case for AI

11:17

in performance reviews when it is

11:19

data focused, when it's like your

11:21

goal was to write 50 articles

11:23

a month rather than me going

11:25

in counting and seeing how many

11:27

articles I wrote a month to

11:29

see if I reach that goal.

11:31

We use AI to like do

11:33

that kind of number analysis and

11:35

data analysis. Are there tools or

11:37

use cases in that way? And

11:39

is that a good use of

11:41

AI for performance reviews? You're beginning

11:43

to see the easy ones, the

11:45

two easy ones are engagement, you

11:47

know, 360 degrees. an engagement kind

11:49

of data about the individual being

11:51

brought into performance management systems to

11:53

complement the narrative. We're beginning to

11:55

see operational data like sales performance

11:57

data coming in. Hey, you made

11:59

your numbers or you didn't make

12:01

your numbers and here's how being

12:03

incorporated. It's still pretty minor. That's

12:05

pretty new because the integrations that

12:07

have to be done to those

12:09

data sources is still very much

12:11

evolving. The agentic AI capabilities that

12:13

are popping up now will improve

12:16

that. But as of right now,

12:18

that's still pretty nascent. It is

12:20

promising, right to your point. Today,

12:22

if I've got to do that,

12:24

there are a lot of managers

12:26

that just don't. And then those

12:28

that do have to do a

12:30

ton of legwork to have good

12:32

data. to serve as the basis.

12:34

So that is clearly a place

12:36

where you would expect to see

12:38

efficiencies. The magic next step, though,

12:40

of turning that into a robust,

12:42

meaningful review is still, you know,

12:44

the human is still that manager,

12:46

right? I don't see any indications

12:48

of automated performance ratings happening in

12:50

any kind of scale. We see

12:52

point examples, right? So if you're

12:54

a coder. you've got code quality

12:56

tests that can creep their way

12:58

into these kind of reviews but

13:00

that's still pretty niche. Yeah it

13:02

does seem like the best use

13:04

case because as you say a

13:06

lot of managers just kind of

13:08

go on for lack of a

13:10

better word vibes you know like

13:12

oh I think you're meeting your

13:14

expectations I'm not going to do

13:16

the extenuous legwork of figuring it

13:18

out so I have a general

13:20

feeling that you're meeting your expectations

13:22

I'm going to rank you at

13:24

meet expectations rather than have the

13:26

hard data that you can't argue

13:28

with, like, yes, you met your

13:30

goals, no, you didn't meet your

13:32

goals, you exceeded your goals by

13:34

exactly this much. We've talked so

13:36

far a lot from the manager

13:38

perspective. From the employee perspective, I

13:40

can see both intimidating to think

13:42

of it this way and maybe

13:44

promising, depending on maybe your relationship

13:46

with your manager. How do you

13:48

think employees should be thinking about?

13:50

this change? So it's a great

13:52

area first off for an employee

13:55

to look to their employer for

13:57

transparency right it's not a bad

13:59

question to ask how is that

14:01

organization using AI in the performance

14:03

review process that shouldn't be a

14:05

secret because this is already a

14:07

fairly low-trust process for all the

14:09

reasons we talked about so if

14:11

AI is yet another black box

14:13

into this it's not going to

14:15

help that so it's a great

14:17

opportunity for an organization to be

14:19

transparent to gather the messaging that

14:21

is being used to the benefit

14:23

of the employee. But I think

14:25

this is maybe a little away

14:27

from the efficiency side of all

14:29

this. I look at what tools

14:31

are like Nopagelam for the Google

14:33

has, are doing, consuming a report,

14:35

a document, creating an easy to

14:37

understand in the moment, learning asset,

14:39

like a short-form podcast, as Nopagelam

14:41

does an example, and using that

14:43

as a way to make the

14:45

performance review more consumable. by the

14:47

individual, not relying on that PDF

14:49

and, you know, sometimes quite fairly,

14:51

we say, PDF on the head,

14:53

give that employee an asset that

14:55

not only helps them understand what

14:57

a three is and how it

14:59

relates to the mission and their

15:01

job, you know, really use it

15:03

as a way to give a

15:05

different way to consume the performance

15:07

review in addition to, maybe as

15:09

pre-work or follow-up to the manager

15:11

conversation. Google's just introduced the ability

15:13

to actually interact with that. They

15:15

bring real-time voice back into that

15:17

equation. So you can ask your

15:19

performance report a question in between

15:21

your conversations with your manager, again,

15:23

not replacing that manager. So I

15:25

think there's a lot of potential

15:27

if we think differently about how

15:29

do we make this valuable to

15:32

the colleague, to the employee. There's

15:34

some really interesting things that we

15:36

can do with not a lot

15:38

of engineering to... Not have the

15:40

oh it's it's the performance review

15:42

time and I'm going to be

15:44

a 3.2 and I won't know

15:46

why Actually creating a tool out

15:48

of it that the person has

15:50

a better chance to be able

15:52

to use and actually develop from

15:54

it. And that's what excites me.

15:56

Yeah, you know, relatedly something that

15:58

we talk about a lot when

16:00

we talk about AI and also

16:02

we talk about a lot when

16:04

we talk about performance reviews is

16:06

bias and human bias and both

16:08

the pros that like does AI

16:10

perpetuate the bias, does it solve

16:12

against it? I will share with

16:14

you and our listeners that back

16:16

in 2014, one of the first

16:18

articles that I wrote for Fast

16:20

Company that went viral. The headline

16:22

was, this is the one word

16:24

men never see in their performance

16:26

reviews, and the word was abrasive.

16:28

Humans in performance reviews obviously are

16:30

open to a lot of bias.

16:32

There's potential there for AI to

16:34

perpetuate it, but there's potential to,

16:36

for AI to perpetuate it, but

16:38

there's potential to, for AI to

16:40

flag those sorts of things. What's

16:42

your experience with how AI can

16:44

help or hinder bias review? Jenny,

16:46

I should do perfectly fine on

16:48

that. And he already is with

16:50

the drafting tools that are out

16:52

there. We're seeing them make suggestions,

16:54

hey, we're this a little differently.

16:56

And so from that perspective, I

16:58

think that's a good thing and

17:00

that's a benefit. But that is

17:02

surface. If you ask me the

17:04

question that's on everybody's mind, these

17:06

models are only as good as

17:09

the data they're trained on, if

17:11

the data has bias, so does

17:13

the model, this use case is

17:15

not going to solve for that

17:17

obviously. So the need for the

17:19

human review, the need for the

17:21

HR professional review, that's one of

17:23

the reasons why I don't think

17:25

it's going away. Because that's going

17:27

to show up more subtly in

17:29

the review text than the George

17:31

Carlin words I shouldn't say, right?

17:33

And I don't think there's a

17:35

magic answer to that quite yet.

17:37

You would hope that's something that

17:39

makes its appearance as the model,

17:41

underlying frontier and foundation models continue

17:43

to improve. but they still require

17:45

an oversight and an implementation of.

17:47

of testing and architecture and monitoring

17:49

tools around the core application, which

17:51

is absolutely essential in this case.

17:53

I want to ask about the

17:55

role of tech in AI-powered tools,

17:57

not just in performance reviews, but

17:59

also in PIP's performance improvement plans,

18:01

such as much dreaded, and in

18:03

layoffs, both meta and Microsoft announced

18:05

recently that they're conducting layoffs based

18:07

on. poor performance, meta even gave

18:09

it that wonderful jargon name, non-regrettable

18:11

attrition. On the surface, it appears

18:13

the fairest way to make cuts,

18:15

if you have to make cuts,

18:17

but it also assumes that there's

18:19

kind of well-defined goals in equal

18:21

metrics for everyone in a clear

18:23

understanding of what good performances versus

18:25

what poor performance looks like. And,

18:27

you know, as we've talked about,

18:29

that is not... always the case

18:31

or frequently not the case at

18:33

most companies. What are the pros

18:35

and cons of using AI tools

18:37

to help make those sorts of

18:39

decisions? Who's an underperforming employee? It'll

18:41

mirror a little bit but only

18:43

to appoint the discussion around the

18:45

review itself as a tool that

18:48

augments the aggregation of data that

18:50

you are going to as the

18:52

human used to make that employment

18:54

decision. Sure, subject to the accuracy

18:56

of the data and the accuracy

18:58

of the interpretation, right, by the

19:00

AI tool. So do I support

19:02

bringing in if you're evaluating the

19:04

performance of the developer, subjective team

19:06

performance 360 degrees for reviews from

19:08

his team and coding statistics about

19:10

how that individual is actually coding

19:12

and bringing that together using generative

19:14

AI as a draft to have

19:16

a hypothesis? Okay, yeah, I can

19:18

certainly go with that. Are we

19:20

anywhere near that being a filter

19:22

or a cutoff or a decision

19:24

point? Not that I'm aware of,

19:26

and I think that would be

19:28

a dangerous place to be. You

19:30

can see a word. If there

19:32

are roles in an organization that

19:34

are so clear on, just as

19:36

you said, so clear on the

19:38

metrics, the KPIs, by the way,

19:40

that probably can't be too many

19:42

of them, that are easily measured.

19:44

Did I make a number? Did

19:46

I make a number? Did I

19:48

not make a number? Did I

19:50

have a certain defect rate if

19:52

I'm a coder or not? Did

19:54

I write the thing, the items

19:56

I was supposed to put into

19:58

the world, or did I not?

20:00

Do I need AI in the

20:02

mix? What's AI bringing in the

20:04

mix, right? Yeah. You know, again,

20:06

I don't see organizations using AI

20:08

to identify poor performing salespeople. They

20:10

look in their CRM, right? You

20:12

can sense in my them if

20:14

we're just beginning to use them

20:16

effectively as information gatherers in a

20:18

one-point performance review. Again, where's the

20:20

plot? I think we're a ways

20:22

away from... using it as an

20:25

objective criteria to eliminate and having

20:27

that be the one and only

20:29

decision poor. We do get it,

20:31

we get us, hey, is AI

20:33

going to fire me? It's a

20:35

valid question and a valid worry.

20:37

Yeah, I think the more that

20:39

we see these sorts of headlines

20:41

and the more that AI becomes

20:43

such a big part of our

20:45

working lives, it does become a

20:47

real fear. And something I keep

20:49

going back to and I talked

20:51

about when I talked about AI

20:53

and hiring is just the nuance

20:55

that is missed and I think

20:57

when it's strictly database. and we

20:59

saw some of the meta employees

21:01

are posting on LinkedIn saying you

21:03

know actually I was a good

21:05

performer or I was on parental

21:07

leave and that's why I didn't

21:09

make my numbers or you know

21:11

all of these very human things

21:13

that happen you definitely need a

21:15

human in there to help understand

21:17

them rather than being judged solely

21:19

as a number. I mean the

21:21

reality is the vast vast majority

21:23

of our clients today will make

21:25

that as a subjective that as

21:27

a one of the most challenging

21:29

difficult impactful leadership decisions that have

21:31

to be made period full stop

21:33

and they are made by humans

21:35

they are almost never black and

21:37

white. They are almost never so

21:39

cut and dried that you can

21:41

just go by the objective data,

21:43

right? Just as you said, there's

21:45

a constellation of reality that informs

21:47

the way a person performs in

21:49

their job that goes far beyond

21:51

their objective metrics. So it's just

21:53

not a space that should, in

21:55

our opinion, lend it to being

21:57

delegated or outsourced to generative AI.

21:59

I don't think any ever, but

22:02

we'll see how this goes. So

22:04

speaking of, when a business approach

22:06

you for a guidance on AI

22:08

and performance management, what are their

22:10

hopes and how do you advise

22:12

them? We get a lot of

22:14

questions right along the lines we're

22:16

discussing. Right now, usually from the

22:18

operational efficiency perspective, right, because managers

22:20

don't like to do it, the

22:22

systems that do it aren't great,

22:24

the feedback is usually pretty obtuse

22:26

and... delivering it is usually a

22:28

challenge. Other than that the process

22:30

is amazing and incredibly well received

22:32

by the participants. We try to

22:34

again get our clients and work

22:36

with our clients to pull back

22:38

a bit and get at the

22:40

transformation that employees are looking to

22:42

make to move along in their

22:44

career and how does that align

22:46

with what the organization needs to

22:48

achieve from the talent to get

22:50

to their business strategy. When we

22:52

do that effectively, the conversation expands

22:54

from how do I make it

22:56

as sufficient as I can be,

22:58

that's still part of it. But

23:00

begins to talk about how do

23:02

we help the leader or the

23:04

manager deliver it more effectively, how

23:06

do we coach these managers at

23:08

scale, but sometimes using AI, to

23:10

be a better feedback giver. And

23:12

then from an individual perspective, what

23:14

do you do with it? Great,

23:16

you got it, you know. But

23:18

a lot of organizations are giving

23:20

ratings without development guidance still, right?

23:22

So you try and bring those

23:24

together. A little bit trying to...

23:26

Move the conversation away from what

23:28

your organization needs is they need

23:30

to objectively rate their employees and

23:32

move it back to a development

23:34

question. How do you improve the

23:36

capabilities and the performance of your

23:38

workforce? And they are a different

23:41

question, even though performances Emily. Well,

23:43

and I think, you know, as

23:45

you're saying, this one thing I

23:47

think that we cover a lot,

23:49

that's a way that people approach

23:51

performance reviews that's flawed, is that

23:53

This is sure, I'll do it

23:55

this one time, we'll get it

23:57

done with, and then we won't

23:59

think about it at all until

24:01

we have to do it again

24:03

next year. And performance management is

24:05

supposed to be an ongoing conversation.

24:07

It's supposed to be, especially with

24:09

career advancement and opportunities that comes

24:11

in along the way. Are there

24:13

effective ways or are there AI

24:15

tools that can help with that

24:17

kind of engagement tools, things that

24:19

kind of go along throughout the

24:21

year and other than this one

24:23

process at this one point? challenge

24:25

and answer enabled by technology has

24:27

been a thing for a while

24:29

now, right? So moving from annual

24:31

performance reviews or quarterly performance reviews

24:33

to something that's more in the

24:35

moment, that's as close to in

24:37

real time as, you know, an

24:39

organization can feasibly make it, that

24:41

has led to kind of a

24:43

coalescing of traditional performance management systems

24:45

and employee engagement tools. That's been

24:47

a technology-driven initiative for some time,

24:49

and it introduces a lot more

24:51

data into the equation when it's

24:53

done well. It can also just

24:55

be yet another thing for a

24:57

manager to do. So now they're

24:59

not only doing the annual performance

25:01

review, they're doing periodic 360s, and

25:03

they're doing project-based assessments. And again,

25:05

it doesn't make the process any

25:07

easier. It adds more touch points

25:09

if it's done well. So you

25:11

would expect and beginning to see...

25:13

AI tools to make that easier,

25:15

as I said, easier to get

25:18

that data together. As an engagement

25:20

stops being surveys and starts being

25:22

more pulses, that's another thing that's

25:24

been going on for a while,

25:26

make it easier to say how

25:28

somebody's doing. You can see AI

25:30

beginning to make that process more

25:32

efficient. We still believe that unless

25:34

you then still consider, does it

25:36

actually help the person? Or are

25:38

you just gathering more data? Fundamentally,

25:40

the reception or the septivity of

25:42

the organization to the performance management

25:44

process is not going to approve.

25:46

And objectively, you probably won't be

25:48

able to look at the process

25:50

overall in the aggregate as an

25:52

organization and say it's making any

25:54

more of an impact just because

25:56

I've added more touch points. That

25:58

rubicross is not cross-cut. So we've

26:00

talked a lot about both the

26:02

pros and the cons and you

26:04

have a little bit alluded to

26:06

where things could go wrong. I'd

26:08

like to end though with a

26:10

worst case scenario and your most

26:12

optimistic scenario for where all of

26:14

this the use of increased technology

26:16

and particularly AI in performance reviews,

26:18

performance management and layoffs in those

26:20

decisions. What's your worst case scenario?

26:22

What's your best case scenario? as

26:24

it applies to performance management, but

26:26

just broadly, are introducing some of

26:28

the most profound leadership decisions organizations

26:30

are going to have to make

26:32

over the next several years, right?

26:34

This is the Pandora's box is

26:36

open, the cover's been ripped off,

26:38

it's been thrown down the hall.

26:40

This isn't changing, this is here.

26:42

So I think any worst case

26:44

scenario starts with leaders in an

26:46

organization outsourcing the implications of AI

26:48

to AI, right? No more. felt

26:50

more obtusely in this kind of

26:52

process, right? So the worst case

26:55

for me absolutely is organizations that

26:57

believe that AI is going to

26:59

take this burden off their plate

27:01

and decide who is going to

27:03

be an employee of my company

27:05

and who will not. There's no

27:07

upside I can see in any

27:09

way you choose to look at

27:11

the problem. So clearly is a

27:13

worst case. Similar worst cases is

27:15

a... that it's a one-sided equation,

27:17

right? That this makes it incredibly

27:19

efficient for the organization. and we

27:21

lose yet another opportunity to make

27:23

this meaningful, relevant, actionable to the

27:25

employee because then trust won't go

27:27

up and we have the opportunity

27:29

to really change the way feedback

27:31

is given, give the human superpowers,

27:33

give them the data and the

27:35

robust feedback tools. make it pre-read,

27:37

make it part of the process

27:39

so that when you are interacting

27:41

with your manager or your reviewer,

27:43

you are having that human conversation

27:45

and not spending your time, well,

27:47

what does a three mean, right?

27:49

I think that's the opportunity here

27:51

to really make it a tool

27:53

for development and advancement of the

27:55

individual and of the organization that

27:57

is happening in the flow of

27:59

work, is happening in the way

28:01

these people are doing their jobs

28:03

every day. That's the potential. I

28:05

hope the nightmare scenario doesn't happen.

28:07

Yeah, I think if I could

28:09

summarize that for myself, it's that

28:11

the worst case scenario is we

28:13

hand it all over to AI,

28:15

the best case scenario is AI

28:17

helps us fix the parts that

28:19

we don't like and have more

28:21

time to improve the human parts

28:23

that we do need to have

28:25

there. If you knew that the

28:27

performance review was a point, hopefully

28:29

of many, that you're getting meaningful

28:31

feedback and guidance and help moving.

28:34

your career law and help do

28:36

your job better? I'm fairly certain.

28:38

You would look forward to it.

28:40

Or at least, maybe that's too

28:42

optimistic, at least not dread it,

28:44

right? At least not dread it,

28:46

right? Yeah, yeah, yeah, okay, that's

28:48

fair, that's fair. But you got

28:50

to do that. If we do

28:52

anything short of that, you will

28:54

look back in this and go,

28:56

you know, all right, we still

28:58

lost the plot. We haven't really,

29:00

you know, you know, found that,

29:02

I guess. Yeah, it has the

29:04

potential then to become yet another

29:06

management fad that we look back

29:08

on and say like, oh, remember

29:10

when we all tried to outsource

29:12

this to AI and what a

29:14

disaster that was. Yeah. Well, Brian,

29:16

thank you so much for being

29:18

here with me. This gave us.

29:20

a lot to

29:22

think about, especially

29:24

in this hotly

29:26

debated topic. I

29:28

really appreciate your

29:30

input. Thank you

29:32

for inviting me,

29:34

Ms. appreciate your input. Thank you

29:36

for inviting me. It was fun.

29:38

For more on AI and performance

29:40

reviews, check more on

29:42

AI for reviews,

29:44

check out the

29:46

show notes for

29:48

this episode for

29:50

related fast company

29:52

articles. And the next

29:54

episode of The

29:56

New Way We

29:58

Work, I'll talk

30:00

about AI and

30:02

AI rights. rights. sure

30:04

to subscribe to

30:06

The New Way

30:08

We Work Way

30:11

you listen so

30:13

you never miss

30:15

an episode. And

30:17

if you like

30:19

this episode, leave

30:21

us a rating

30:23

and review on

30:25

Apple leave us a rating

30:27

The New Way

30:29

We Work is

30:31

hosted by me,

30:33

Kathleen Davis, We and

30:35

produced by Henry

30:37

by me, Kathleen Nelson, and

30:39

and Joshua by Henry

30:41

Mixing by Nicholas

30:43

Torres. and Joshua Christensen with

30:45

mixing by Nicholas Torres.