TTL on TotalPicture Radio Podcast to Discuss AI Trends Report (Full Transcript)

We sat down with Peter Clayton of TotalPicture Radio to discuss our latest Trends Report V7 on Artificial Intelligence from Candidate Sourcing to Engagement and Selection.

.@TalentTechLabs President, Brian Delle Donne, discusses the Vol. 7 Trends Report with TotalPicture radio: Click To Tweet


This spring, TotalPicture Radio will be reporting from and covering a number of the most important and influential talent acquisition, HR, leadership, and technology conferences, including SourceCon in Anaheim, California, The ERE Spring in beautiful San Diego, California, TAtech Spring Congress co-locating this year with SHRM Talent Management in Chicago, Illinois.  To learn more about these events visit the TotalPicture Radio conference and events page on For more information regarding our unique, strategic, custom sponsorship opportunities, and video production services send an email to with media kit in the subject line; or give us a call at 203-293-7003. And now, here’s our show.

A short time ago I came across a Talent Tech Lab’s trends report that clearly spells out the common misconceptions and hype associated with how artificial intelligence (AI) and machine learning is about to disrupt everything, including Talent Acquisition.  I reached out to Jonathan Kestenbaum (@JKestenbaum), the Executive Director of Talent Tech Labs who introduced me to the author of the report and President of Talent Tech Labs, Brian Delle Donne.

A little history. Talent Tech Labs was originally founded by the New York City-based staffing firm Mitchell Martin, Inc. to foster the growth and development of emerging ideas and companies in the talent acquisition technology space. The company is an incubator and accelerator that mentors startups on site at 307 West 38th Street in New York and through their virtual incubation program.  

Welcome to an innovation channel podcast on TotalPicture radio. I’m your host Peter Clayton.

In our interview, you will hear an educated and research perspective on AI, machine learning, deep learning, and neural networks specifically as these technologies relate to and support Talent Acquisition. Now, here’s my interview with the President of Talent Tech Labs, Brian Delle Donne.

Peter: Brian, in my intro I gave the boilerplate introduction to Talent Tech Labs. Perhaps you could provide us with a little more information and background.

Brian: Sure, Peter. Talent Tech Labs is an innovation center that started off by trying to promote the improvement of the recruitment process all across with being associated with recruitment through the application of emerging technology. That’s taken several forms.  

In the early days, we incubated early stage companies. We still do that but, over the course of time we have programs for later stage companies that develop technology in the space, as well as information services for people that buy and evaluate these technologies. So, corporate Talent Acquisition executives, people in staffing,  and recruitment process outsourcing companies, as well as, strategic participants in the ecosystem that have adjacent technologies that may be interested in extending their own product reach or functionality.

We’re a small team but we work very closely. We’re entirely focused on Talent Acquisition Technology which is a subset of the HR tech field.

Peter: Tell us a little bit about your background, Brian.

Brian: Actually, I was trained as an engineer but never practiced as one, and grew up through a number of work experiences early on in sales and project development and then eventually in leading companies – usually companies that were delivering professional services. I ran an environmental services company which was an engineering firm. I worked in the staffing industry. I’ve worked in the software implementation industry. Most recently, it was a boutique staffing company that was very innovative in how they tried to apply technology that had the brain child of starting Talent Tech Labs. That happened about four years ago while I was here as the chief operating officer of a staffing company called Mitchell Martin. The last four years have been entirely focused on building Talent Tech Labs.

Get your copy of the Trends Report Vol. 7, right here:

Trends Report MockupPeter: I want to start out talking about this Talent Tech Labs trends report.  Listeners, you will find a link where you can get this report on Brian’s show page. I want to start out with a quote that’s in this report. It’s from Harvard Business Review.

“The truth is AI is as “fully automated” as the Great and Powerful Oz was in that famous scene from the classic film where Dorothy and friends realized that the great wizard is simply a man manically pulling levers from behind a curtain. This blend of AI in humans who follow through when the AI falls short, isn’t going away anytime soon.  Indeed, the creation of human tasks in the wake of technical advancements has been a part of automation’s history since the invention of the machine lathe.”

Brian, give us some background and texture into how you approached developing this very, very interesting and thoughtful report on machine learning and AI.

Brian: We’ve been speaking on this subject for quite a while but we felt compelled to come out with something that was as definitive as this in part because there’s a lot of confusion around the term, the terminology, and the actual presence in our workspace today, and particular in recruitment. I guess the quote that you cited was an illustration of our thinking that says yes, artificial intelligence is here, it’s expanding, it’s getting better but, there is and continues to be a large component of human involvement in how the machines work.  

There has been a bit of paranoia at least in the recruitment space, that recruiter jobs are going to be taken over by machines and so, a bit of hysteria seems to have arisen in the talk circles. This report in part was trying to just calibrate folks so that they had a better appreciation of where we are in this progression. In my assessment we’re kind of in the first going on second inning, so we’re very early in the game.

Peter: To that whole thing about paranoia, back to my friend Jeremy Roberts who’s speaking at SourceCon this month, his talk is titled, “How to Ensure You Survive The Machine Learning and Artificial Intelligence Disruption.” So, you are absolutely right; a lot of people think the bots are coming and they’re going to take my job.

Brian: You know, I think if you look at the way things get automated over the course of time, the process tends to take root and it has in the case of recruitment around things that are highly repetitive, somewhat mundane, don’t require human skills, particularly. Those are the first things to get automated. One of the things that we try to explain in this article is that automation alone is not Artificial Intelligence.  

You see people dropping the AI term with broad, broad strokes when in fact, much of what’s been accomplished heretofore has really just been sort of smart automation of prophecies like workflows, or events that are triggered through some activity based function. So the idea that these constitute Artificial Intelligence I think is a wild overstatement.  

As my piece in this report dives into, there’s actually a number of components that make up the field of Artificial Intelligence, one of which is machine learning. For machines to actually learn it requires several things to be present, not the least of which is large bodies of data against which machines can be trained to learn.

When you’re talking about things in recruitment largely unstructured data, we’re still a ways away before machine learning can actually capture and start to deduce higher level understandings from this body of data. Not to say that it can’t over time, but right now the human component of recruitment is really front and center, and we think that it’s going to be augmented by the use of AI to leave more time for the human elements to actually play a role and eliminate from the human engagement those things that are highly repetitive and can be programmed and handled by machines.

Peter: A subset of this machine learning is natural language processing (NLP). Can you define what that really is and how that relates to recruiting?

Brian: Natural language processing has a number of applications in the recruitment space in part because languages are complex and we are describing things both in job descriptions and in résumés and in cover letters, and company branding statements that use language. For machines to start to process these kinds of contextual aspects of spoken language is a complex thing and so when you apply it, it can reduce quite a bit of time that’s necessary to digest information.

There are a number of things that need to be learned. In particular, context and where machines are getting good at that, but in the area of understanding intent that is still a very, very distant objective to be solved through machine learning.

Peter: Talk to us a little bit about deep learning, which I guess is the next level up from machine learning.

Brian: Deep learning is modeled on the way that the human mind actually thinks. In a sense, it’s a way to automate predictive analytics. When we take an input as humans it forms a layer of our thinking and then we get another input and it modifies the inputs that you previously received. Over the course of time we build up a complex level of abstractions in a hierarchy, if you will, is what leads to knowledge. In deep learning, it’s really stacking algorithms one on top of each other in these hierarchies each of increasing complexity and abstraction that sort of mimic the way the human brain works.

Programs that do deep learning go through iteration upon iteration until the outputs that come out actually arrive at what we would consider maybe an acceptable level of accuracy. But the interesting thing in machine learning is that at the early stages it’s learning what the programmer wants us to see and is trying to solve for a particular outcome. But once these things get really smart and you have enough layers of abstraction, machines can posit outputs that humans might not have been able to get to simply because they have been able to crunch through so much more data that would be impossible for a single mind to comprehend on its own. So, machines ultimately can be more predictive through recognizing patterns in the data and inferences as they crunch through massive amounts of input.

Peter: Then you get into talking about neural networks, the cloud, and big data. Can you explain this to our listeners please?

Brian: If you think about the comments I just made that for machines to learn they need to be able to sort through massive amounts of data. Those data sets could exist in single places in the past, main frames inside of large corporations like insurance companies or banks or payroll systems have massive amounts of data that is in one place. However, the rise of neural networks which is actually a combination of software and hardware that result and can be visualized as a sort of the cloud of computing allows this network to be formed of data sources that reside in many different places across the planet. Even in the ether.   

In the past, you might be reliant on from the data that your company’s captured internally as your only point of comparison. Now with neural networks, the data is widely available coming from multiple sources; some for sale, some for deduction by yourself by interpreting information that you’re  able to deduce from behavioral reviews of products to the way people decide what kind of movies to watch. The data is being drawn across a network of machines and infrastructure that allow these massive data sets to actually be tapped into for machines to learn against.

Peter: As you know, Brian, there are a lot of companies out there now that are using predictive analytics to help assess who is a flight risk within a corporation, for instance, and who out there would be a good fit within a specific role within a company. How do you see all of this advancing and maturing over the next couple of years?

Brian: As I said, the process that’s iterative and as more data gets exposed to these kinds of algorithms, I do think that they will be trued up and become more and more accurately predictive. We’ve been studying at Talent Tech Labs we see all the early stage companies coming to market. We’ve seen pitches from people that have reported to have this kind of capability two or three years ago. It occurred to us that there is quite a bit of aspiration in those statements as opposed to reality. Because although they can aspire to say, hey, these algorithms will eventually be able to be good predictors of who’s going to stay or who’s going to be promotable, or who’s the best fit; the truth of the matter is in the time those statements were made two or three years ago the data sets were not available for them to have educated machines or algorithms to be able to come up with those kind of findings with any accuracy.

So very scant data sets and you were probably drawing conclusions that were maybe happenstance circumstance as opposed to being a correlation or a causation.  Over time as more behavioral data is crunched, as more career trajectories are mapped, as more detail that could be exposed to these algorithms to come into play the picture will become more clear as to who’s a better fit, who’s more likely to stay or go, or be promoted.

Peter: In your research report you have a vendor’s perspective who is Shon Burton who is the CEO of HiringSolved. What did he contribute to this report and what is his approach when it comes to artificial intelligence?

Brian: Shon’s an absolute genius. He has been a very big contributor to the space.  He has been working on software that is making matches and filtering by making matches. As AI has grown in availability and prominence he has been a user of these tools to perfect the offerings that he has brought to market, not all of which are based on artificial intelligence but certainly as AI is applicable or has become applicable he’s implemented them.

He now in a current offering that comes from their company, they have a bot, a chatbot that is able to engage with candidates and able to gather information and use it to refine searches and make better matches. He’s very widely read and a very respected expert in the field.

Peter: There’s a chart in your research report that’s called Sophistication of AI Technologies. It goes through all kinds of things like Watson, Tesla Auto Pilot, Alpha Go, Seri, Cortina, and Amazon’s Alexa. I guess what Shon is really working on is a recruiting version of what you would consider to be Alexa. Right?

Brian: It’s more of a chatbot, so the chatbot would be able to interpret language obviously, and then formulate responses, ask subsequent questions based on the input from the last question asked. So, it’s not a linear set of questions that are fired off at the respondent without being first recalibrated to the answers they gave on the previous question. It is combination of chatbot and then the voice recognition and natural language processing.

Peter: What did you learn in researching and writing this report that perhaps surprised you or was unexpected?

Brian: It never ceases to amaze me at how the human mind continues to perfect and work things. I guess what was a little surprising is just how far back in time some of the basic science that goes into machine learning has been in play. There have been elements in the scientific and academic communities that have laid the groundwork for this over the period of 30, 40 years so, we’re only seeing a current manifestation. Some of the processes, the thought processes, the exploration into how the mind works and the formation of algorithms is actually rooted in some pretty good work but not necessarily current. I’m amazed at how long this has been going on.  

Now we’re seeing a really rapid acceleration. I guess that’s always amazing to the Moore’s Law Concept as to how quickly things can multiply.  We are going at a very rapid pace by comparison to the history of the subject.  

Peter: That’s an interesting insight there of the whole Morris Law comparison because at the point the exponential increase in the amount of data and the processing power is really driving all of this advancement.

Brian: That’s true. That’s true. The availability of the neural networks has really been sort of a watershed event for unleashing vast quantities of data.  Now we’re at a point and this has been probably for more than 10 years, people recognize the value of data and whole businesses have built around just the capture of it and then the repurposing of it. So, you’ve got a couple of things happening at once. People recognizing the importance and the value that’s embedded in that data and then the infrastructure becoming available now where it can be relatively easily accessed.

Peter: What are some of the questions you get from companies coming into your organization and saying ‘hey, we’re really looking for an application or something that can do X.’ What’s the X?

Brian: I think in the talent acquisition space, which is really all we focus on, they come in and they site a pain point. From that pain point they don’t necessarily know what’s going to solve that or help them alleviate that. We hear challenges that range from – and over time they’ve evolved, right? At one point people were saying we can’t source enough of the kinds of people that we need to fill our pipeline. And then more recently as sourcing tools have become more prevalent and effective, people were saying I’ve really got to figure how do I engage. I have all these people that are sending me responses.  I see all these qualified candidates. They seem to be qualified but how do I get to engage with them. How do I get them to want to take my call?  How do I cut through the noise? So, engagement types of questions.

Another extreme situation is talent acquisition leadership team telling me that they’ve been running a tremendous branding program so that their company is attracting some of the best and brightest résumés out but  they’ve got stacks of thousands of them they’ve not reviewed yet and they’re feeling remised that they continue to spend on driving advertising budgets to promote their brand, all the while not really digesting even the information they have right on their desktops. How do you help me solve the fact that I have captured all this stuff now, how do I get down to filtering it the ones that I really need to talk to.  

We hear these points of business pain as the way they particularly what’s challenging for them and then we try to point them in the direction of certain technology or applications of technology that we think most closely address that.

Peter: Well, you know, it’s really nice to hear. I’ve heard this from other resources as well that the candidate experience is really something that the companies are starting to focus on. You’re absolutely right, Brian; the companies are spending thousands and millions of dollars on employer branding and doing all kinds of wonderful things with their Facebook fan pages, commercials, and all kinds of advertising things and then the candidate gets to the career portal and goes to their applicant tracking system and everything just sort of blows up.

Brian: Yeah, it’s very discouraging. That’s an interesting observation though because as we see start-ups and founders coming in with new ideas;  I can’t tell you how many have become entrepreneurial because of bad experiences in the hiring process. There are some really, really smart people coming out of first class schools or out of first class corporations, management consultant firms, people that have very good credentials saying, ‘geez, I’m ideally situated for this job. How come they didn’t even give me a call? How come they didn’t even acknowledge?’ And that, in some cases, has led them to just stop what they’re doing and try to design a solution to improve what they perceive that challenge to be.  

Some of the innovation is actually coming as a result of these experiences which are pretty daunting if you’re trying to get heard through a technology layer that just hasn’t been designed to be consumer centric, if you will. You know, that applicant tracking system, it’s history has been to be a system of record and largely to keep corporations compliant with things like EEOC and requirements that are driven by federal mandate or corporate mandates.  They’ve got a function that is nowhere near candidate engagement. It’s really about compliance.  

It shouldn’t be any surprise that a system that’s been designed for compliance purposes is going to have a pretty unappealing interface when pointed at a candidate experience.

Peter: To your point about engagement, I’ve heard this from a lot recruiters and companies as well that it’s a huge concern.  I recently had Craig Fisher from CA Technologies, he is the head of employer brand there on the show.  One of their strategies has been to build talent communities where they can have these candidates become part of a talent community and start a conversation with them so they just don’t lose them.  I think a lot of companies are really starting to see the creation of a talent community as a way of capturing and engaging qualified candidates that perhaps to your point where they can’t keep up with all the submissions they’re getting, at least they can start a conversation.

Brian: Yeah, Craig has some very good insights into the employer brand.  He’s one of the leading proponents of corporations trying to work on that interface.  In cases where the companies take the time to try and improve that is had materially positive improvement in Net Promoter Scores and in candidate engagement. There’s no doubt that it works if it’s done effectively.  It’s not an easy task. If you’ve heard Craig, and he’s probably got into it on your show, there are many facets that need to be addressed concurrently; it’s not just hey go and put a few engaging little blog posts out there that might be interesting to this community. It actually does revolve around content but as Craig told you, the content could be from the very interpersonal level of the employees that are inside the company becoming advocates to the company, to other ways to get the corporate culture to be more prevalent to the applicants than would traditionally have been the case.

Peter: You’re right. It takes money and commitment and people to do this.  It’s just back to this whole thing about there’s not some magic AI thing or bot out there that is going to do this work for you. It takes real people to do the work.

Brian: It does and you know what Craig would have said in one of his –  I’ve seen many of his presentations. It’s not a theory what he’s witnessed and helped orchestrate at CA, he’s turning the employees of the place into brand  ambassadors. That’s kind of a cheap way to do it. If you had to go out and hire advertising people or writers to come in and create this persona there’s also the risk of it becoming artificial or not being accurate. But, empowering employees to become brand ambassadors – Starbucks is another great example of this; you talk to the people that have worked there that represent that brand and they’re all about how great a place it is and why.  

So by empowering the employees, unleashing the employees and the alumni to be your ambassadors is a tremendous way to increase brand awareness. If there’s a way to connect applicants would put the job might be like on the other side I think you would have a more engaged and positive applicant pool if they have those awarenesses. The ones that have paid attention to understand that will in fact, have sort of self-selected to be a smaller set of those that would probably have a better chance of succeeding at that employer.

Peter: Right. One of the things Craig told me is that the CEO of CA is 100 percent behind this, and they have a very open and transparent culture there. That’s what drives this kind of brand ambassador within the organization. The culture has to align with that type of attitude.  

Brian: That’s absolutely correct. You can’t make that up. It’s got to be sort of grassroots and really be present for it to show through as such.

Peter: Brian, I really appreciate you taking time to speak with us here on TotalPicture Radio. Is there anything that we haven’t covered in this conversation regarding your research report that you’d like to share with the audience?  

Brian: That’s a big question. I think what we’re trying to say is artificial intelligence and machine learning is prevalent in our lives today. It is going to continue to make improvements in the way talent acquisition works but it is not going to be necessarily at the expense of human involvement. We posit that it’s going to actually require for humans to apply different skill sets that they’re equipped with which are really those of evaluation skills, empathy, actual verbal communication.

We think that it’s an exciting time. Not one that should be inspiring paranoia but it should be actually helping people get excited about the possibility of how we can be doing a better job in recruitment with these new technologies coming to add to our efforts.

Peter: How can our listeners connect with you and with Talent Tech Labs?

Brian:  I’m Brian @TalentTechLabs. My website is  We’re very collaborative. We have quite a few followers. If you simply come to our website and download any piece of information you’ll be in our database and start to get our periodic publications. There are trend reports. There’s the news ticker that comes out monthly. There’s an events ticker which talks about all the events in the talent acquisition conference space that talk to the various events that new thoughts are being presented at and vendors are presenting at as a way just to network and stay better in touch with the community.  

Peter: Are you going to be at any of the Spring Conferences, SHRM, ERE, Source Con, or any of these places?

Brian: Yeah, many. We divide and conquer here because there’s just too many for all of us. Jonathan Kestenbaum our managing director and I share time between these things. In addition to the ones you’ve mentioned, The Human Capital Institute has got a conference in June which is on talent acquisition. Canadian Staffing Association has a conference in April that’s focused on talent acquisition. We have staffing industry analysts coming up with a thing called the Gig Economy in the September timeframe. HR Tech World Congress is actually coming to the United States for the first time this year in June in San Francisco.  So, there’s a much more complete listing of these things.  Along with meet-ups and much smaller local events that we sponsor or call out on our website.  

Hopefully, our listeners today if so inspired, would look us up and avail themselves to some of those resources.

Peter: Great. Brian, I look forward to meeting you in person at one of the conferences coming up.

Brian: Thanks, Peter. It’s been great talking to you today.

Peter: Thank you.

Brian Delle Donne is President of Talent Tech Labs.  You’ll find our interview with many resource links in the innovation channel of TotalPicture Radio.  That’s  Thank you for tuning in today.  We sincerely appreciate your participation.  Your comments are welcome on Brian’s show page.  While there, please sign up for our free newsletter, subscribe to our show on iTunes, Google Play, or Sound Cloud and join the conversation on our TotalPicture Radio Facebook Group.  You’ll find me on Twitter @PeterClayton, @TotalPicture, and @Jobsinpods.

Be sure to let me know if you’ll be at Source Con this month, ERE Spring, TAtech, or SHRM Talent Management this April. I’m looking forward to seeing you and extending the conversation.  

One more thing before you go. LEVER, a next generation ATS based in San Francisco is hiring. Check out my job cast with LEVER sales recruiter, Michael Gallagher on where real employers talk about their jobs and tell you how to get them.

Trends Report: Mastering Remote Work Management