Design Training That Builds Real Business Tools with Rose Benedicks

1 - > Jay Johnson: Welcome to this episode of the Talent Forge, 2 - > where together we are shaping workforce behavior. 3 - > I am excited to bring in a fellow ATD member. 4 - > For those of you that don't know, the ATD is the Association 5 - > for Talent Development. 6 - > And I got to meet Rose at a conference not all that long 7 - > ago.

8 - > So I am excited to bring her in here to have some beautiful 9 - > conversations. 10 - > Rose and I have a lot of shared knowledge in some different 11 - > spaces, and I am excited for you to learn from her today. 12 - > So welcome to the show, Rose. 13 - > Glad to have you.

14 - > Rose Benedicks: Thank you very much, Jay. 15 - > I am really excited to have an open conversation with you and 16 - > see what interesting things we get into. 17 - > Jay Johnson: Yes, definitely. 18 - > So, Rose, tell me first, how did you get into this space of 19 - > talent development?

20 - > Like what was your interest? 21 - > What brought you in there? 22 - > And sort of what excites you about it? 23 - > Rose Benedicks: Well, the the easier question is what excites 24 - > me about it.

25 - > And honestly, uh, it's twofold. 26 - > I absolutely love that you just get to learn random stuff. 27 - > You never know what kind of skills you're going to be 28 - > developing, what kind of behaviors you're going to be 29 - > developing. 30 - > And it leads you down these paths of having to learn fairly 31 - > quickly all kinds of interesting things from gosh, um, physics 32 - > lessons and pneumatic lessons for Marines driving 33 - > expeditionary fighting vehicles to uh sales negotiation and 34 - > consulting skills and relationship building.

35 - > So you never know where it's gonna go and what you're gonna 36 - > get to learn. 37 - > And that is fantastic. 38 - > As far as how I got into it, I am one of those rare birds who 39 - > always wanted to do it. 40 - > When I was a kid, I used to come home from school, redesign 41 - > the lessons, and teach them to my stuffed animals.

42 - > So I was always gonna do this. 43 - > Jay Johnson: I love that. 44 - > And so often we hear on the show, oh, I fell into it 45 - > accidentally. 46 - > I did this, and then I got there.

47 - > So it's really interesting to hear the perspective of wanting 48 - > to get into this space. 49 - > And I'm gonna say I do echo one of my favorite pieces is this 50 - > week I'm working with engineers, next week I'm working with 51 - > doctors, the week after. 52 - > So I get to play a little of all of those on TV. 53 - > That's actually one of the parts that really fascinates me 54 - > too.

55 - > So tell me more about this, though. 56 - > You come home and redesign the lessons. 57 - > Was it because there was like a sense of like I can do this 58 - > better? 59 - > Or was it one of those of like that didn't land for how how did 60 - > how did your brain quantify wanting to do that?

61 - > What did that look like? 62 - > Rose Benedicks: Gosh, well, my nine-year-old brain, I think, 63 - > just was excited about having learned something and was 64 - > turning it over in my head, going, oh, here's a way it makes 65 - > sense. 66 - > Let's let's try this. 67 - > Here's a way for me to share my knowledge.

68 - > I was a pretty precocious child, and I um half of my 69 - > family are professors. 70 - > So, you know, that that wanting to share knowledge and and, you 71 - > know, quite frankly, probably talk about it more than I 72 - > should, uh, was always in there, always driven by example. 73 - > So I think it was just my way of sort of making sense and 74 - > sharing, which has always been my MO for learning something. 75 - > If I can share it with somebody, I'll know it better.

76 - > Jay Johnson: I love that. 77 - > And it's so true because one of the things that I have found, 78 - > and it is well documented amongst the research and 79 - > learning, is when you start teaching something or when you 80 - > start uh sharing that knowledge with somebody else, whether it's 81 - > the the drive to bond that sort of like creates that shared 82 - > connection, or that's the actual, hey, if I have to teach 83 - > this, I better know it a little bit, whatever that aspect is, we 84 - > do tend to remember, recall, and also retain more of the 85 - > information.

86 - > So for those of you out there that are learning something, go 87 - > teach somebody immediately afterwards. 88 - > You are highly likely to actually retain it, be able to 89 - > recall it, and be able to use it. 90 - > So, Rose, you have a lot of knowledge and experience that 91 - > you're able to retain, recall, and use. 92 - > Talk to us a little bit, and and I want to kind of start.

93 - > You know, one of the things so long ago, when I it wasn't all 94 - > that long ago, but so long ago when I got to see you speak at 95 - > the ATD conference, you had such a great presentation. 96 - > I thought it was so fascinating, and it was a very 97 - > novel way to do it. 98 - > It connected with me immediately with some of my past 99 - > experiences. 100 - > Um, do you recall which one I'm talking about?

101 - > I don't want to put you on the spot here. 102 - > Rose Benedicks: I'm pretty sure it was on workplace outputs as 103 - > the goal of training. 104 - > Jay Johnson: Yes. 105 - > Can you speak to that concept?

106 - > Because I think that the way that you framed that was 107 - > probably one of my favorite talks in a long time because it 108 - > was all about outputs, getting people to actually have a return 109 - > on investment for the dollars that they're putting into 110 - > training, which is something that is so important and 111 - > something I've been a big advocate for. 112 - > So the way you framed that, the way you shape that, I thought 113 - > was really powerful. 114 - > And there should have been a hundred CEOs sitting in that 115 - > room listening to this and saying, this is where we should 116 - > be putting our dollars.

117 - > Talk to us a little bit about that. 118 - > Rose Benedicks: Well, gosh, that's high praise. 119 - > So let me see if I can talk as eloquently as your compliments. 120 - > Um let me give an example first, because you know, I can 121 - > talk about all the benefits and talk about it in these 122 - > highfalutin terms, but let's let's land the plane to begin 123 - > with.

124 - > A really simple example would be uh, let's say you're 125 - > onboarding technical trainers, facilitators who are going to 126 - > teach people how to maintain HVAC systems, for example. 127 - > Onboarding might traditionally look like here are the tools, 128 - > here are the systems, go explore the equipment, try some 129 - > troubleshooting. 130 - > But what if for those uh trainers, onboarding, the output 131 - > of onboarding is you have created your facilitator guide.

132 - > Right. 133 - > Jay Johnson: Gosh, I love that. 134 - > And and the the funny thing is, is when you say it, people are 135 - > like, huh. 136 - > Well, that seems like that could be logical.

137 - > You're actually creating something that's going to be a 138 - > business value beyond just training somebody, but actually 139 - > going through the structure and building that. 140 - > But nobody does this. 141 - > Help help us understand why why is it that most trainings, most 142 - > trainings, most talent development does not come into 143 - > place with creating something like an output? 144 - > Like, here's the facilitator guide for future use and use 145 - > cases or future onboarding, et cetera.

146 - > Why what is it that stops us from getting to that level? 147 - > Rose Benedicks: Well, it's really hard, quite frankly. 148 - > Uh so you know, when we think about learning and performance, 149 - > we're often pinpointing a performance issue or a specific 150 - > need. 151 - > We people are are X and we need them to be able to do or 152 - > produce Y.

153 - > And you know, it's easy to default to knowledge, we all 154 - > know that. 155 - > And then it's really, you know, we have this great hype around 156 - > scenarios and story-based learning. 157 - > Let's actually have them practice in a safe space. 158 - > But I think it's a lot harder to say, let's practice in the 159 - > real space.

160 - > And by the way, it's not practice, you're actually doing 161 - > it. 162 - > So a great example for those who are maybe more 163 - > executive-minded would be this concept of human in the loop 164 - > when using AI. 165 - > So let's say you've got an AI agent that will flag potential 166 - > fraud for a credit card company. 167 - > The human in the loop says before they freeze that account, 168 - > somebody has to review it.

169 - > Right? 170 - > So you've got to train all these people in the workplace 171 - > now to work with AI. 172 - > So, what does that look like? 173 - > How do I come up with my plan for reviewing that flag?

174 - > And if the if instead of training on here's how you 175 - > prompt the AI, here's how you um make a decision, here's the 176 - > algorithm for it, here's the decision tree. 177 - > Instead, you train people on having to make the correct 178 - > decision. 179 - > The output is the correct decision. 180 - > Now, what's cool about this and why it's so hard is because you 181 - > have to get into the way of doing work, right?

182 - > So you're not giving people a program on how to interact with 183 - > AI, you're building interacting with AI into creating that 184 - > output. 185 - > So you're flipping it on its end. 186 - > So if you are in a company where people need to 187 - > collaborate, like let's say you're creating a high-volume uh 188 - > patient plan for a um ER or emergency department, if you're 189 - > creating that or updating that, you've got to talk to a lot of 190 - > people. 191 - > You've got to talk to the nurse manager, you've got to talk to 192 - > the hospital admin, you've got to talk to data analysts.

193 - > Um, there's lots of steps and whatnot. 194 - > Instead of training somebody on here's who you talk to, here's 195 - > what the plan needs to include, you create the plan. 196 - > And the blended learning includes who do I talk to about 197 - > this? 198 - > When do I talk to them?

199 - > What inputs do I need? 200 - > And you're actually then imbuing the way of working into 201 - > getting that output. 202 - > Jay Johnson: I love that phrase, the way of working. 203 - > And and I think about this because traditional training, 204 - > traditional talent development.

205 - > I go to a conference room, I sit there, I listen to somebody 206 - > tell me what I should do, how I should do it, and why maybe, 207 - > maybe, maybe I'm gonna say why I should be doing it that way. 208 - > And I say maybe because we would think that that would be a 209 - > part of it at this point in time, but it's not always. 210 - > So we get often the how and the what. 211 - > We don't always get the why, sometimes we do.

212 - > And then we leave there, and then the expectation is just 213 - > that we can just, you know, monkey see monkey do. 214 - > I just learned it, so now I can go do it and actually the way 215 - > of working, I think, is such the novel concept. 216 - > And for those of you listening in, I really want you to think 217 - > about this because the outputs that come from these trainings 218 - > are often the return on investment for spending money on 219 - > trainings. 220 - > And I've seen way too many LD departments getting cut, getting 221 - > slashed, people losing positions because, well, things 222 - > are uncertain, economies are uncertain.

223 - > And one of the first things to go is talent development. 224 - > And that's really sad. 225 - > If you want to elevate your status as a strategic business 226 - > partner and not a luxury item inside of your organization, 227 - > you've got to figure out how to impact the way of working and 228 - > the outputs that are coming, that is going to be I put $1,000 229 - > in and I save myself $1,500 later. 230 - > If we're not thinking like that, more than likely your 231 - > talent development, your change management programs are going to 232 - > get cut.

233 - > Is that a fair statement, Rose? 234 - > Is that your experience? 235 - > What are you looking at when you're implementing some of this 236 - > type of training? 237 - > Rose Benedicks: Yeah, absolutely.

238 - > I think you said something really smart there, Jay. 239 - > So this concept of ROI, and yes, as an industry, we have 240 - > been talking about it forever and ever and ever. 241 - > But here's something really interesting. 242 - > If the outcome of your training is an actual value-added 243 - > output, workplace output for the company, two things happened in 244 - > terms of ROI.

245 - > One is you can be a little more confident in your correlation. 246 - > That training, the output of it, was a work product. 247 - > We can therefore better correlate that that training 248 - > resulted in something of value, right? 249 - > That disconnect is it gets a little gets a little shorter.

250 - > You can follow the chain of evidence more confidently. 251 - > And then there's the obvious one, which is yes, you have a 252 - > return on investment because you created something of value for 253 - > the organization, right? 254 - > You didn't create knowledge, you created an actual thing 255 - > that's going to be used upline or downline, and you created it 256 - > in a way that reinforces how you want your company to work that 257 - > built fundamental skills along the way, like how to interact 258 - > with AI, how to work across departments, how to break down 259 - > silos.

260 - > So there's, you know, there's a really big advantage here in 261 - > terms of getting the bang for your buck. 262 - > Jay Johnson: And I think that's huge. 263 - > And we're going to step back into this AI in just a moment, 264 - > but can you give some other examples of maybe some of the 265 - > outputs? 266 - > Like broaden the audience's mind.

267 - > You you had mentioned, you know, obviously the HVAC one, 268 - > but what are some other examples of outputs that could be 269 - > created through the learning experience or training 270 - > experience that would be a value proposition to any 271 - > organization? 272 - > What are your thoughts on that? 273 - > Even if it's just a kind of a short list. 274 - > Rose Benedicks: Gosh, I think, you know, risk management plans, 275 - > right?

276 - > Creating a risk management plan, keeping it updated, 277 - > business continuity plans, absolutely. 278 - > Uh performance management plans, right? 279 - > So the output would be you actually have your plan for how 280 - > you're going to manage performance. 281 - > You know, that's aligned with how the organization wants to do 282 - > it.

283 - > Uh, sales enablement is a big one. 284 - > So that could be anything from um, you know, better using you 285 - > know AI to better predict which leads are likely to close, and 286 - > coming up with you know, my plan for how I'm gonna follow up on 287 - > that, right? 288 - > Actually, have my plan for making those database decisions, 289 - > for making those predictive decisions, um, creating 290 - > marketing campaigns. 291 - > How do I you know the output of that might be the actual 292 - > campaign itself?

293 - > It might be the the um agent, the agentic workflow in the CRM 294 - > is the output, right? 295 - > And you're and you're actually creating the output through the 296 - > training. 297 - > And the tricky part then is you design the training to create 298 - > the output, not just step one, get some data. 299 - > Step two, make a decision, but it's you know, step one is you 300 - > have to get the data.

301 - > Here's how to look at the data. 302 - > Go talk to this person who's going to help, you know, 303 - > understand this part if that kind of collaboration is 304 - > important to your organization. 305 - > And then you have natural checkpoints along the way, 306 - > right? 307 - > Because we're not just talking about the workplace output full 308 - > stop, we're talking about the workplace output to a specific 309 - > set of standards and requirements, right?

310 - > And there's no better way to evaluate somebody's work than by 311 - > looking at the real product of it. 312 - > The biggest barrier, I know I'm kind of going on a path here, 313 - > but the biggest barrier I see is clients will tell me people 314 - > don't have time to look at other people's work. 315 - > And I say that is a big fat lie. 316 - > Because they're either gonna look at it or they're gonna redo 317 - > it, and that's the same amount of time.

318 - > Jay Johnson: Yep. 319 - > Or they're gonna look at it under crisis, right? 320 - > And you can either do this now, and and it's funny you say 321 - > that, like one of the, and I didn't know what this was called 322 - > and or or even the concept of it, but with one of the very 323 - > early trainings that I developed. 324 - > So part of my educational background was in crisis 325 - > management and crisis communications.

326 - > And one of my very early trainings was essentially the 327 - > output was here's your crisis communication plan and here's 328 - > the different strings. 329 - > But it was teaching them like, okay, this is how people think 330 - > in this stage of the crisis. 331 - > What would we need to do? 332 - > How would we need to do this?

333 - > And then we would literally develop like, here's the first 334 - > uh communication, here's the first communication template 335 - > that we're gonna send out. 336 - > It's gonna do these four things. 337 - > Let's put the information in there. 338 - > And by the time that we would finish the program, they 339 - > literally would have a start to front, like open the book, step 340 - > one, here's where we need to be.

341 - > Step two, this is who we need to contact. 342 - > Step three, this is how we get everybody on the same page. 343 - > Step four, this is how we send out the first piece. 344 - > So the training was teaching them not only the psychological 345 - > mindsets of themselves, the organization, the audiences, et 346 - > cetera, but it was also about them building something that in 347 - > the event of a crisis, they've got a plan.

348 - > And yeah, so you know, when you were sharing, uh when you were 349 - > sharing your presentation, I was just like, it was one of those 350 - > things where I was like, this is so brilliant. 351 - > I accidentally, and like after your session, I'm like, here's 352 - > the nine things that I would do differently now that I've 353 - > learned, uh, to kind of form uh make that a little bit more um 354 - > formatted and even, you know, sort of uh, I'm gonna call it uh 355 - > uh a little bit more polished, a little bit more toned.

356 - > But uh I thought it was such a cool concept because I 357 - > accidentally kind of stumbled into that just because I felt 358 - > everyone should have a crisis plan. 359 - > Almost nobody did. 360 - > So it was one of those things where it's like, all right, this 361 - > is what you're gonna get. 362 - > But putting the value on that, and it's funny because I have 363 - > had clients come back and say, we had X happen, we had this 364 - > crisis plan, no one panicked, and it was and and it worked 365 - > except for usually it was a except for this one part.

366 - > And it's like, okay, well, we can train to that part now, but 367 - > you're out the plan. 368 - > So I I just really loved that. 369 - > Um, so thank you for sharing that with the ATD, and thank you 370 - > for sharing it with the audience here today. 371 - > Audience, be thinking about this.

372 - > When you're enacting some level of talent development, what is 373 - > the output? 374 - > How are you going to actually make a long-term organizational 375 - > difference? 376 - > How are you going to help your teams to have something? 377 - > This is a really cool model.

378 - > And Rose will share her uh contact if you ever want to 379 - > reach out to her about some of these different things later on 380 - > in this episode. 381 - > I want to go back to AI. 382 - > I know you've been doing some pretty cool uh study and work in 383 - > this space. 384 - > Tell us a little bit about what is your experience and you 385 - > know, sort of what has you excited about that?

386 - > Rose Benedicks: Oh, well, there are again, I'm gonna go with the 387 - > with the two answers. 388 - > From an LD perspective, I think LD is positioned in a really 389 - > interesting space to help figure out what working with a digital 390 - > coworker looks like. 391 - > So LND has long grappled with sort of elusive capabilities. 392 - > What is critical thinking?

393 - > What is resilience, right? 394 - > What are those things? 395 - > What is agility? 396 - > How do we pin that down?

397 - > What does that look like in action? 398 - > And now we're going to be faced with taking some of those more 399 - > elusive behaviors and pinning them down, creating 400 - > capabilities. 401 - > What does it look like to work with an AI or digital coworker? 402 - > Uh, what does it look like to be a human in the loop?

403 - > That's very different. 404 - > And L and D can get into a space where they're helping to 405 - > find that because we do have those, we do have those set of 406 - > competencies as an industry. 407 - > The other part about AI. 408 - > So right now I'm in an executive ed program for AI 409 - > transformation and leadership.

410 - > And what I'm finding most fascinating is how future-built 411 - > companies are really doing it. 412 - > It's a co-ownership model between the business and IT. 413 - > There's FinOps at play, you've got to get the data structure 414 - > right, you've, you know, you've got to get an interoperable tech 415 - > stack. 416 - > There's a lot of base requirements.

417 - > And everything that I am learning, all of the research, 418 - > everything from the big four consulting firms, all says your 419 - > talent strategy better be in there. 420 - > Jay Johnson: Yeah. 421 - > I I can see that that would, and unfortunately, I mean what 422 - > is your experience? 423 - > And I'm gonna ask this.

424 - > I've I've got at least some experience. 425 - > What is your experience with organizations and or businesses 426 - > being ahead of the curve when it comes to AI? 427 - > Rose Benedicks: Oh, what a great question. 428 - > So there is a framework where you look at organizations in 429 - > terms of their AI adoption from stagnant, lagging, emerging, and 430 - > future-built.

431 - > And if you look at those future-built organizations, they 432 - > have an AI operating model. 433 - > They're not using AI for productivity, which I think is 434 - > where a lot of people get stymied. 435 - > So using it for productivity might look like you know, 436 - > generating content, segmenting your customers more quickly. 437 - > But when I say using it to create value, I'm talking about 438 - > a return that shows up for shareholders.

439 - > Now, what those companies are doing differently is they're not 440 - > they're starting with the fundamental strategy, a clear 441 - > strategy and vision. 442 - > Co-ownership, like I said, between the business and IT, uh, 443 - > they have an actual investment plan with ROI clearly stated for 444 - > how they're going to structure AI. 445 - > They mandate near-term gains when you're talking about going 446 - > from your central infrastructure out into departments and how 447 - > they're using it.

448 - > Near term, they're mandating near term gains. 449 - > Um, and then, like I said, the talent strategy that seems 450 - > scary, I think, to a lot of small and mid companies because 451 - > the investment seems so high to get the fundamental platform, 452 - > hybrid cloud, whatever you want to call it, ego. 453 - > Systems set up. 454 - > But when we talk about value, uh, think about a supply chain 455 - > and agent-to-agent ecosystems, which is where one agent can 456 - > talk to another agent and do stuff on its own, essentially.

457 - > So think about supply chain. 458 - > Something breaks down in supply chain, a supplier's not going 459 - > to fall, come through. 460 - > Your AI agent can go talk to other suppliers' agents, get, 461 - > you know, negotiate an alternate uh avenue, an alternate 462 - > supplier, figure it out, get it all done, and by the time your 463 - > people come in, they have an email that says this supplier 464 - > wasn't able to meet demand for whatever reason. 465 - > We negotiated, it's now coming from these people, it'll be here 466 - > in this time.

467 - > Right? 468 - > That's what we talk about with value that shows up for the 469 - > shareholders. 470 - > Jay Johnson: All while somebody's sleeping. 471 - > Now, take the human in the loop concept, because I know that 472 - > that would be I know that there are some people, whether it's a 473 - > manager or a leader out there that's listening, that's going, 474 - > that's terrifying.

475 - > What if it hallucinates? 476 - > What if it makes this mistake? 477 - > What if it, you know, uh, you know, I I still want to rely on 478 - > my purchasing team to be doing the negotiations. 479 - > Like, what's what's the answer to that?

480 - > Because I there are some legitimate fears out there. 481 - > Don't get me wrong. 482 - > Uh, you know, I just recently asked uh ChatGPT, and it, you 483 - > know, I was having a little bit of fun with it. 484 - > I was like, what about this song by this band?

485 - > And it came back and it was like, that song doesn't exist by 486 - > this band. 487 - > I'm like, I'm literally looking at my iTunes and I'm like, that 488 - > song, that band. 489 - > I'm like, you're hallucinating. 490 - > Please look again.

491 - > And of course it comes back. 492 - > So we've all experienced something, whether it's a small 493 - > hiccup or whether it sourced some information from Reddit and 494 - > shouldn't have happened. 495 - > How do how do we get sort of the attitudinal change? 496 - > And that's really where I think that there is some aspects of 497 - > that attitude can come from fear, that attitude can come 498 - > from this is taking away as opposed to uh leveraging growth, 499 - > et cetera.

500 - > How do we get past that attitude barrier that may be 501 - > holding some companies back? 502 - > Rose Benedicks: Well, that's a big barrier. 503 - > Uh, first of all, let me apologize for putting everyone 504 - > to sleep, and I'm gonna try to make this a bit more 505 - > interesting. 506 - > So, you know, there's a concept that, you know, generative AI, 507 - > predictive AI, agentic AI is this big thing and it's trained 508 - > on everything.

509 - > But if you're gonna do it well for your organization, you're 510 - > gonna pick the precise model for the tasks at hand. 511 - > So when you're having that kind of precision strike, you do 512 - > get, you know, you can get a better handle on hallucinations. 513 - > And if you have a virtuous cycle of training that 514 - > particular model in its specific use case, then you also can 515 - > avoid model drift, right? 516 - > Um, where your training becomes sort of outdated.

517 - > And I'm talking about the AI training data, it becomes 518 - > outdated, and so your model drifts. 519 - > And those two are the big ones AI hallucination and and model 520 - > drift. 521 - > Now that's a really key concept. 522 - > You deploy the precise amount of compute power needed for each 523 - > task, and you have strict governance, so you're not kind 524 - > of going out there into the gooey mass of everything.

525 - > You have a fundamental, you know, AI infrastructure that has 526 - > data and whatnot, but then you go out into the departments and 527 - > the specific tasks and use cases get the computing power they 528 - > need. 529 - > Just doesn't go big and get everything. 530 - > And in doing that, first of all, it's a huge cost savings. 531 - > And second of all, the governance becomes a lot easier 532 - > to put into place.

533 - > Jay Johnson: And I I want to piggyback on that because it was 534 - > actually interesting to me. 535 - > And this is audience. 536 - > If you're using AI and not training your AI, you're making 537 - > a mistake. 538 - > And and and what do I mean by training that training the AI?

539 - > And I thought that this was um, and it's really interesting 540 - > because I kind of I've I've seen AI used where it's a search 541 - > engine. 542 - > All right. 543 - > It's basically the new Google. 544 - > I've seen it used at some very advanced levels.

545 - > One of my colleagues, one of our certified guides and 546 - > behavioral elements, is actually literally building out some 547 - > incredible models, uh, spent hundreds of hours of time 548 - > training it uh to actually do like personalized coaching 549 - > towards uh managing health outcomes with uh pre-diabetes 550 - > and things like that. 551 - > So like really advanced things, really cool stuff. 552 - > Um the AI, teaching the AI all of the knowledge related to 553 - > different medications, to different pieces uh to be 554 - > looking for, inputs, outputs, all of it.

555 - > Really, really cool. 556 - > I think in some cases, one of the fastest pieces that people 557 - > can start doing is teaching its AI what it's allowed to look at 558 - > and what it's not. 559 - > So, audience, here's a quick thing. 560 - > And and again, just did a talk on this um when in the ATD 561 - > behavioral science thing.

562 - > And it was interesting because I was like, how many of you are 563 - > actually telling your AI, you are only allowed to look at 564 - > these resources from Harvard business, from this, from this, 565 - > from this. 566 - > You are not allowed to look at Reddit. 567 - > You are not allowed to look at this, you are not allowed to 568 - > look at this. 569 - > Now help me come up with a strategy or a plan for X, Y, and 570 - > Z.

571 - > And they were like, you can do that. 572 - > And it was just like, okay, yes, this is what you should be 573 - > doing. 574 - > That training piece, I think, is really fascinating because 575 - > people don't realize like AI is not a one-way communication 576 - > model. 577 - > It should be a two-way where it's iterative and back and 578 - > forth.

579 - > So, are do you have any tips on training AI? 580 - > You know, from the individual level, not even necessarily the 581 - > corporate or the enterprise level. 582 - > Are there things that you have done to make sure that your AI 583 - > is more dialed in than say just the regular, I've gone to chat 584 - > and opened up a new, uh, you know, opened up a new new talk? 585 - > What what are your thoughts on that?

586 - > Rose Benedicks: Yes. 587 - > So, you know, I do some of what you said, don't look at this. 588 - > Right. 589 - > Um, I've also had a conversation with uh my AI of 590 - > choice, uh, which is Claude, um, about the kinds of sources that 591 - > I view as credible and not.

592 - > And then I've had it sort of save that search and recall it, 593 - > right? 594 - > So maybe in a later prompt, I'm telling it, hey, remember that 595 - > chat we had, put that into play here. 596 - > Um I also will ask it to, if it ever cites anything, give me a 597 - > link so I can go verify it myself. 598 - > And then, you know, that's when I'm using a broad um gen AI 599 - > model, right?

600 - > But there are specific models that even individuals can use 601 - > now. 602 - > So for example, a quad cowork, which is sort of a productivity 603 - > tool that um integrates with your personal computer, and it's 604 - > not out there trained on everything that a generative AI 605 - > large language model is trained on, right? 606 - > It's looking at your data. 607 - > It's going, what's spam, what's not, what kind of emails do you 608 - > want to respond to first thing in the morning?

609 - > What should I bubble up? 610 - > How do I, you know, tell you, you know, hey, you're gonna need 611 - > some time to prepare for this meeting based on the notes from 612 - > your from your last call, these are your action items, right? 613 - > That's not out there being trained on everything. 614 - > That's actually a really fun space to play around because 615 - > it's smaller and you can start telling it, this is important to 616 - > me, this isn't.

617 - > This is valuable, and you're right about this, but you're 618 - > not. 619 - > So getting in and playing around with those kinds of 620 - > models is really helpful versus trying to tackle wrangling in a 621 - > huge model like Chat GPT or Claude or whichever one you want 622 - > to use. 623 - > Jay Johnson: I love it. 624 - > And I I'm gonna I'm gonna share this.

625 - > And we've had enough conversations, you probably 626 - > figured out that I'm a little bit of a nerd when it comes to 627 - > some of the tech, some of the science, and particularly 628 - > behavioral science, but there is a mixture between feeling like 629 - > a, you know, an adoration for Tony Stark, uh Iron Man. 630 - > Yes. 631 - > So I have taught my AI different protocols. 632 - > So I have, you know, something like Research Protocol One, 633 - > which is the highest level research protocol.

634 - > It knows it is only allowed to look at peer-reviewed journals, 635 - > particularly I have an entire list of them if it's doing this 636 - > type of thing. 637 - > Research protocol two is like you're allowed to now dip into 638 - > things like Forbes, Wall Street Journal. 639 - > Research protocol three is you're allowed to look into 640 - > something like psychology today or some of the things where it's 641 - > not all peer-reviewed. 642 - > There's articles, there's guest contributions, et cetera.

643 - > So I've set those up. 644 - > I do also have Christmas protocol, um, which is like, 645 - > hey, I need to actually do my Christmas shopping. 646 - > So I've got a number of different protocols, but I've 647 - > trained it to know when we dip into a particular protocol, 648 - > here's the parameters that you're allowed to do, here's the 649 - > things I don't want you to do. 650 - > And I just have little names for them so that way it makes 651 - > sense for me that I can say, hey, we're doing research 652 - > protocol one, jump in there, here's what I'm looking for, 653 - > find me 15 articles that say X.

654 - > And then I've got, you know, a nice list or let's synthesize 655 - > this. 656 - > Let's find what are the patterns that we're seeking. 657 - > But it knows how to look at those things because the first 658 - > several times it didn't work. 659 - > So then it was like, okay, you did this, that was smart.

660 - > I want you to continue to do this. 661 - > I want you to stop doing this, I want you to uh expand on this 662 - > type of thing. 663 - > We don't think about training the machine, we think about the 664 - > machine training us too often. 665 - > And I think it's really important that general users 666 - > start really looking at it as that two-way.

667 - > Rose Benedicks: Yeah, and this goes right back to our primary 668 - > topic of workplace outputs. 669 - > Um that when you're creating something at work, and we're all 670 - > going to be using AI to do it, right? 671 - > So, what you just talked about, how do I make the AI better? 672 - > And not just the large language model that that our company has 673 - > adopted, but in this workflow, when it has told me, hey, you 674 - > need to, you know, this is a human in the loop moment.

675 - > Um are you able to make it better? 676 - > Are you able right? 677 - > So in in even if you're creating a marketing campaign, 678 - > there's a moment where you have to learn through creating that 679 - > output to make the AI better at giving you what you need. 680 - > And that becomes part of the training that gets you to that 681 - > output.

682 - > So instead of attacking it and going, you need to learn how to 683 - > make your AI work better for you and have these ideas for you 684 - > know, workflow design and whatever, it's now part of the 685 - > training to get to that output. 686 - > Jay Johnson: Love that. 687 - > So, with this, with this executive ed program you're 688 - > you're working in in the AI and leadership space, how do you see 689 - > yourself working with organizations? 690 - > Like if you were to say, hey, my you know, my ideal here is I 691 - > want to help organizations do what?

692 - > What does that look like for you? 693 - > Because I know that your internal is the teacher. 694 - > So you're gonna be teaching this and and and moving that 695 - > forward. 696 - > What does that look like in that space?

697 - > Rose Benedicks: I really want to join an organization where my 698 - > enjoyment of creating a framework or an approach out of 699 - > complexity and chaos is gonna come into play. 700 - > Where my desire and sort of natural inclination to create a 701 - > team that's quite resourceful and share some enthusiasm to get 702 - > that done comes into play. 703 - > So it's more about the again back to ways of working and the 704 - > way my brain works. 705 - > And I want to work in a space where that's valuable, but also 706 - > where I'm gonna learn, which is the whole reason why I'm doing 707 - > this AI program.

708 - > It's not because I'm out there going, I need to be this, you 709 - > know, AI great person, otherwise we're you know, I'm never gonna 710 - > get a job. 711 - > It's like, wow, I'm very curious. 712 - > And AI is such a systems concept, so it kind of works 713 - > with my my desire to see systems, to improve the internal 714 - > logic of a system to make it better. 715 - > Um, so I'm leaning a little bit towards the operations space, 716 - > more strategic operations.

717 - > Um, and you know, AI is just gonna be a part of that moving 718 - > forward. 719 - > Now, having said all of this, let me be very clear. 720 - > I am not a technical person. 721 - > So, big disclaimer for anything I've said.

722 - > Um, I just give really good face when I'm with tech 723 - > geniuses. 724 - > Jay Johnson: No, I I love that, Rose. 725 - > And the best part is is so the first time that I spoke on AI, I 726 - > was uh I did an international conference all on AI. 727 - > I'm the only non-tech person there, literally the only 728 - > non-tech person there.

729 - > Um, I know enough to be dangerous, but I am certainly 730 - > I'm not a coder. 731 - > I'm not somebody who is somebody that can build 732 - > something. 733 - > I understand some of the background, but not all of it. 734 - > Uh, but I understand people.

735 - > And in order to get the technology to work, you have to 736 - > have the people component and that human in the loop. 737 - > So usually when I'm brought in on something, I'm with you. 738 - > I'm not a tech person. 739 - > I dabble, I like it, it's fun.

740 - > Um, but that human side, and and the thing that I think 741 - > really resonated with me, with what you said, is the tech is a 742 - > tool. 743 - > And if you don't have a system in place to manage the tool and 744 - > the interactions with it, all you're going to do is create 745 - > more chaos. 746 - > And I see a lot of companies doing that now by not creating 747 - > the systems for how do we interact with AI? 748 - > How do we create this as a future program or future 749 - > opportunity?

750 - > So I'm going to ask you kind of one question before we close 751 - > out our conversation. 752 - > First of many, I hope. 753 - > Um, but I the question that I would have is if if if I'm in 754 - > the audience and you were to say, I want to start, you know, 755 - > and I was to say, I want to start moving to sort of that 756 - > futuring of AI. 757 - > What is maybe one of the first things that I should be thinking 758 - > about, acting on, or doing to move me more from the I'm using 759 - > it as a search engine or I'm using it for efficiency, as you 760 - > said, to that long-term.

761 - > What is something that I should be thinking about doing 762 - > differently or even just starting in order to get there? 763 - > Rose Benedicks: Yeah, so wow, gosh, there are so many starting 764 - > points. 765 - > I would say look at your workflow end to end, not a 766 - > single part, but end to end. 767 - > And what are the big questions?

768 - > So, for example, in HR, well, let's go L and D. 769 - > That's our audience, right? 770 - > L and D and HR. 771 - > So you're looking at yeah, you're looking at the life cycle 772 - > of an employee.

773 - > So you start with how do we attract the talent? 774 - > And how do we bring them in? 775 - > How do we get them going? 776 - > How do we keep them going?

777 - > How do we make them happy? 778 - > So you look at that entire workflow end-to-end and go, how 779 - > can we reimagine this? 780 - > If if we want to meet a person at every step of the way and 781 - > have them feel a certain way or experience a certain thing, how 782 - > do we reimagine that workflow? 783 - > That's where you start.

784 - > Because when you get into, hey, how do we make our hiring 785 - > decisions faster? 786 - > You're gonna have all of these little different AI-inspired 787 - > projects and pilots and experiments. 788 - > Your resources get spread too thin, it's wasted money, and 789 - > nothing is coalescing to create actual value. 790 - > You've just created little pockets of efficiency.

791 - > Jay Johnson: Okay. 792 - > I'm gonna selfishly, I'm gonna say I love that because we just 793 - > started that process in uh our behavioral elements team of 794 - > looking at the entire workflow of uh recruiting new guides, um, 795 - > onboarding them, helping them reach their first, you know, 796 - > first sets of uh training success and coaching success to 797 - > building out their enterprise and everything else. 798 - > And we're looking end to end on this one. 799 - > And we've been utilizing some AI in there to help us.

800 - > What are our gaps? 801 - > Where is somebody going to get a hiccup? 802 - > What is their experience? 803 - > So even looking at like attitude and looking at 804 - > enjoyment and frustration models.

805 - > So you said this and I was like, we're doing it right. 806 - > So yeah, no, that's very cool. 807 - > Rose, if our audience wanted to get in touch with you and tap 808 - > into the wealth of knowledge and experience that you have in 809 - > this space, how would they re how would they do so? 810 - > Rose Benedicks: Well, I'm on LinkedIn and I'm pretty good at 811 - > checking it.

812 - > However, if you really want to slide up into my messages, uh I 813 - > would just use Rose Benedict's at gmail.com. 814 - > Uh that's uh last name is B-E-N-E-D-I-C-K-S. 815 - > Jay Johnson: Perfect.

816 - > We will make sure that that's in the show notes. 817 - > Rose, thank you for joining us today. 818 - > Like I said, when I had first seen you at that ATD conference, 819 - > I knew I wanted to have a deeper conversation with you 820 - > about the outputs about being able to build a training program 821 - > that's actually creating business value. 822 - > I would I was so enthused and excited about the talk.

823 - > Um, still am, obviously. 824 - > Uh, and I'm hoping that the audience can see the value of 825 - > like, wow, we can do actually more than just knowledge 826 - > transfer, but we can actually build tools, systems, and 827 - > experiences that are going to build our thing forward. 828 - > And I think with the knowledge you're sharing today and the 829 - > excitement that you share it with, uh, it becomes much more 830 - > real. 831 - > So thank you for joining us.

832 - > Rose Benedicks: Well, Jay, thank you so much. 833 - > And I got to tell you, um, you know, I only learn things from 834 - > other people. 835 - > That's my MO. 836 - > So I am really interested in people contacting me because 837 - > they have something to share.

838 - > I spoke a lot, but I also want to learn. 839 - > Jay Johnson: That's amazing. 840 - > So, audience, don't hesitate to reach out to Rose. 841 - > And I'm sure that we're going to have you back because there's 842 - > a lot more to talk about.

843 - > But thank you, audience, for tuning into this episode of the 844 - > Talent Forge, where together we are shaping workforce behavior. 845 - > Thanks again, Rose. 846 - > Rose Benedicks: Thank you.