a musing. @philalbinus - Tumblr Blog

Wy is no one talking about this?!

How Nestle Hires 62K Workers Each Year

As the Global Talent Acquisition Sourcing & Solutions Manager for Nestlé, Guilherme Neves has a mandate that would intimidate any TA professional. Neves serves the global food and drink processing conglomerate’s 500 recruiters in 53 regional markets where each team boasts a head of talent acquisition and varying sizes of TA teams. Each year, they fill 62,000 positions from more than one million unique applications they receive every year. AMS Verified spoke with Neves about his role, how technology serves his global TA teams, and how a customer version of ChatGPT is helping Nestlé hire for the future. This interview has been lightly edited for clarity and brevity.

Please tell us about your responsibilities as Global Talent Acquisition Sourcing & Solutions Manager at Nestlé?

Guilherme Neves: I work in the corporate TA team, which means I report to the global Director of Talent Acquisition and as part of the team, I'm responsible for the strategy, services, and solutions for all Nestlé businesses in all markets. We do not have recruiting needs per se, because my team’s main purpose is to support the TA teams to deliver the best recruitment they can.

What are your TA technology challenges?

Neves: Currently we're using SAP SuccessFactors as our main ATS and it’s the backbone in our TA toolkit, and on top of that, we have additional tools that will help us innovate, source our talent pipeline, and engage with candidates proactively.

Because Nestlé is so large, we operate in nearly 200 countries and our challenge is having a level of localization that is not possible to achieve today. If someone in Australia wants to implement a big change in recruiting, requisitions, or how they post jobs, unfortunately it's not possible because we work on a global scale. Everything is standardized across all markets but it's also a limitation.

What CRM does Nestle use?

Neves: We use Avature as our CRM but only in the U.S. Outside of the U.S., we don't have a particular CRM tool, so we use the features inside of SuccessFactors for the candidate pipeline, engagement, and recruitment marketing. We do use HireVue in some countries, but not globally, and we use Modern Hire depending on the location.

Are you looking into Generative AI tools?

Neves: We have in our IT acquisition roadmap a process of how we can integrate better with artificial intelligence. We are looking at how to build a better integration for automatically generating jobs job posting. Right now, we are building solutions and training our TA community to get them ready to start using and adopting more ChatGPT.

We have our own version of ChatGPT that has passed all the compliance and safety assessments and our IT teams. We call it NesGPT.

Is NesGPT the Nestlé version of ChatGPT or is it a custom model that you've created?

Neves: No, it's a Nestlé model of ChatGPT. It's the same as ChatGPT and the only difference is it went through a compliance review. Our own version is completely secure, and it doesn't capture any Nestlé employees data or any data points that we enter in the system.

Even before we had the NesGPT version, we recommended that the TA teams replace the Nestlé name with “XXX.” Some of the confidential data that would allow someone to identify us as a company, we recommended to replace with something else.

What are some of the specific challenges in hiring in the fast-moving consumer goods (FMCG) space?

Neves: It varies. If you go to countries where we have a higher unemployment rate, the challenge is not that they don't have candidates, they have too many. They need better tools to screen them further. Europe is a different story. In Europe, maybe in the U.S. or a few other countries, we have fewer applications and we need to source the candidates. We need a strategy to understand the types of skills that we need to source candidates from.

There's also a movement where some people believe that it's better to work in small business. They can have a bigger impact in small business instead of having a small impact in a larger organization. Then I would say we are always trying to look for digital types of skills. It's not that we need developers because we buy those products, but we need cybersecurity and data integration analytics. Those skills are difficult to find.

Is there a magic bullet that you wish technology companies would create?

Neves: If I can have only one, the first one is how we keep all the thousands and potentially millions of candidates we have engaged with us. We have so many applications but how can we keep them connected with us? That is one crazy challenge for us. We have over one million applications per year …

Only one million?

Neves: We hire a large amount of people, about 62,000 direct hires a year, but we have more than a million applications every single year. The large majority of those will be rejected because we don't have a million jobs, but how can we keep them engaged with us? They are still people that deserve our respect. How can we keep them engaged and connected with us, the business and the brands?

Please tell me, when are we going to get that product?

philalbinus

RPO & the India Advantage

Long known as the destination for call centers and back-office processing, the outsourcing and offshoring giant is pivoting to R&D, engineering, data analytics and now AI.

Here’s why your organization must join Team India.

In the years since the global pandemic and the subsequent lockdown, India pivoted to become a center of true technological innovation that is dedicated to the next-wave advancements in programming, research and development, engineering, data analysis and the newly introduced business tools: artificial intelligence and Generative AI. India outsourcing centers also embrace hybrid and office-based working models which help broaden the search for candidates who are entertaining multiple job offers in a tight labor market.

Let’s look at the India RPO advantage by the numbers:

- India’s total labor force is an estimated 437.2 million, i.e., larger than the total population of the United States (338.3 million)

- India’s Global Capability Centers employ up to 70% of the world’s GCC headcount

- Last year, India reported an overall leasing volume of 27.3 million sqft in the offshoring industry, a 26% increase from the previous year

- Demand for AI services could reach $17 billion by 2027, according to a report by IT industry body Nasscom

- Last year, around 45% of high-tech and travel companies, and 43% of telecom, manufacturing, and construction firms, nearshored operations to India, per media reports

In a few short years, India is now not only an exporter of high talent, it’s attracting and retaining talent as well. As forward-thinking talent acquisition leaders looking to expand its recruitment process outsourcing, it’s time to prepare for the next India business revolution.

Read the entire article here

philalbinus

EEOC's Sonderling: AI regulation is coming but don't forget the laws that are currently on the books

The EEOC's Keith Sonderling testified before the US Senate today, February 27. Here's a 2022 article I wrote about the commissioner's views on Generative AI.

As the use of Generative AI expands into the HR and talent acquisition spaces, recruiters must not only prepare for new state, federal and multinational regulations that govern hiring and promoting employees with the aid of machine learning, they must continue to comply with employment rules from the 1960s. Using new, ground breaking technology is not a green light for employers to ignore these established laws, warned Keith Sonderling, commissioner for the U.S. Equal Employment Opportunity Commission in a recent webinar. Despite the widespread belief that using AI in HR and TA matters is unregulated because the technology is new, employers must still comply with such employment laws as Title VII of the Civil Rights Act, The Americans with Disabilities Act, The Age Discrimination and Employment Act pertain to any decisions made with the aid of AI.

“With the global discussions about how do we reign in, stop and regulate AI, [people incorrectly believe] that AI is unregulated and they can do what they want, and vendors can sell and create what they want,” said Sonderling.

“And that's just not true,” he added.

If employers use ChatGPT and similar tools to post and advertise job openings, examine resumes, select candidates for interviews, make a job offer with a specific salary, or promote an employee, this is an employment decision that must fall in the letter of the law. “At the end of the day, there is still going to be a decision being made. It hasn't created some new sort of employment decision that we've never seen before,” said Sonderling. “That's really my job here at the EEOC, to remind both the candidates and employees being subject to these tools, the people who are developing or deploying it, [they should] ask, ‘what is that decision that you're asking the tool to make?” he said.

And make no mistake, new regulations are coming.

Along with last year’s Local Law 144 in New York City, recent proposal from the European Union and the various proposals from states such as California, Massachusetts, New and Jersey, HR and talent recruiting leaders will see common themes emerge from these legislations.

“That may be a good indication that that's where these laws are going to go. And that's what employers need to actually start complying with in advance of a national federal law,” said Sonderling.

Sonderling said he and his team at the EEOC have been encouraging employers to conduct a yearly pre-deployment audit and a yearly bias testing audit much like the ones required by the NYC Local Law 144.

“That's the same kind of employment testing for disparate impact that companies have been doing for a long time based upon the EEOC standards from the 1970s,” he said.

Some recruiters are using social search tools with a diversity filter that use databases of candidates and their backgrounds to create a shortlist of diverse talent.

Sonderling agreed that this violates civil rights law. “When you're using the tool to filter out one specific group that you want to hire based on a protected characteristic, this is just as unlawful as firing somebody because of that characteristic.”

EEOC Commissioner: Recruiters must avoid AI recruiting tools that intentionally select candidates based upon their protected characteristics

Thanks to new social search tools that scour social media and other sources, recruiters are now able see or the AI can predict the race, ethnicity, gender, age, or potential disabilities of candidates on so-called “blind” job applications.

As the recruiting world tackles the latest generation of AI tools, the commissioner for the U.S. Equal Employment Opportunity Commission warns that recruiters and their client companies must take special care to not use these advanced tools to discriminate when hiring candidates.

Using AI to source diverse candidates for a company’s applicant pool is helping companies increase the diversity of their applications. However, using AI to hire candidates solely based on their race, gender, ethnicity, religion, or any other protected characteristics in an effort to reach an organization's DEI goals violates long-standing civil rights laws, warns Keith Sonderling, EEOC Commissioner and a former management-side labor and employment lawyer.

It's not a dual-edged sword, he says. “While federal contractors and certain other employers have an obligation to affirmatively seek out diverse candidates or simply want to diversify their applicant pool, in most instances, it becomes unlawful if employers are making the final employment decision based upon such categories,” he says.

“You can use the latest AI technology to source the most diverse applicant pools, but the bottom-line decision on who to hire must be based upon their actual ability to perform the job at hand, through their qualifications, skills, or potential," warns Sonderling. "The AI cannot be used to intentionally to source and hire only based upon immutable factors unrelated to their qualifications such as if they are male or female, African American, White, or Hispanic, or under a certain age."

But a new raft of AI technology can make hiring candidates based on these protected characteristics alluring to harried recruiters who are under immense pressure to hire workers from diverse backgrounds.

Thanks to new social sourcing tools that scour social media, blog posts, online resumes websites, corporate biographies, and other information sources, recruiters are now able “see” or the AI can “predict” (whether directly through photographs or through associational characteristics, like for membership affiliations) the race, ethnicity, gender, age, or potential disabilities of candidates on so-called “blind” job applications.

With this information that previously wasn’t included or discernable from a paper resume, employers can now use the technology to include or dismiss them in the interview process, solely based upon these factors. In some cases, candidates won’t be able to make it through the earliest screening phases of the applications and be dismissed outright in the recruitment process, if the tools are used for such improper purposes.

Recruiters using these tools, for example, can make assumptions based on a resume featuring the name of a historically black university, religious institutions, or the name of a women’s college sports team, even as this information has been used in the past to eliminate candidates with these achievements on their CVs. But for the most part, recruiters cannot use this information for the sole purpose of hiring or not hiring these candidates.

“The problem with these programs is that … [recruiters] actually see what the candidates look like or are provided with a specific breakdown organized by race, gender, national origin or age—in milliseconds for 100,000s of thousands of resumes,” says Sonderling.

Sonderling notes that a more appropriate use for employers who want to use a skills and merit-based approach and comply with their legal obligations not to discriminate is to use AI to strip away direct or unintentional biases. Using software to mask protected characteristics such as gender, age, national origin, and hiring based on a candidate’s skill through AI, is a practice that has gained traction in the post-Covid, quiet-quitting hiring landscape.

For example, he points out that using an AI chatbot or an AI-assisted audio interview, instead of a traditional in-person interview or now a video interview for a candidate’s first round of interviews avoids unintended biases. This prevents recruiters from “seeing” if the candidates belong to what the EEOC calls “protective characteristics” such as those associated with race, gender, disability, religion, and pregnancy.

“You’re not seeing that they're pregnant or disabled but [the AI bot] looks at them based on their merits,” says Sonderling. “Now, you're actually judging candidates on the basis of their skills, not what they look like or where they are from.”

Even relying on frequently used terms by Silicon Valley firms in job advertisements such as seeking a “rockstar coder” or a “programming ninja” might favor male candidates over female prospects. Sonderling says that changing the wording of a job description to favor more gender-neutral terms fosters a more inclusive applicant pool.

“But when you're using AI to say, we want to limit our search to black men who just graduated college between 20-25 or Hispanic females under 30, you're making a decision based upon a protected characteristic as opposed to simply interviewing the most qualified candidates using neutral characteristics like if they meet the minimum requirements of the job posting,” says Sonderling.

There’s a legal and reputational risk as well. Actively using AI to find and eliminate certain candidates could trigger federal investigations and correlating financial penalties from the government and in some cases class action lawsuits. There have been numerous examples of internal emails, for example, from companies that sought to hire a person from the BIPOC community for a specific role that boiled over onto Twitter damaging the organization’s reputation and of course, violating the law.

How are local, state and federal governments able to prove the unlawful use of AI in recruiting? This is where AI makes the recruitment process more transparent, according to Sonderling.

“If you're using AI to unlawfully include certain or exclude certain individuals, let’s say based upon age, now you have a digital record available that will be eventually produced in legal discovery,” he says. “You have ex post facto [proof] of them clicking and making the AI produce and select or exclude these exact candidates.”

Who owns the risk in an AI Anti-Bias audit?

Now that employers must test their AI hiring tools every year, questions remain on who is ultimately responsible if instances of biases have been found.

The New York City law that requires Manhattan-based employers audit any AI tools they use for hiring candidates (as well as promoting employees) for evidence of bias has raised plenty of questions for organizations that are adopting machine learning for their TA operations

Called “NY LL144,” this regulation applies to employers using automated employment decision tools (AEDT) to evaluate candidates or employees who reside in New York City for a position or promotion. They must conduct the bias audit prior to using the AEDT and any job candidate that is a New York City resident must be given notice of the use of the tool, where a candidate is defined as a person that has applied for a specific position by submitting the necessary information or items in the required format.

But if an employer is using an AI-powered talent acquisition solution and the audit finds an instance or multiple instances of bias -- intentional or unintentional -- who owns the risk, the vendor or the employer?

The risk falls on the employer, say leading TA solution providers.

According to Manto Papagianni, Head of Product for TA solution provider and AMS Verified partner Bryq, employers must understand that while vendors have the option to procure bias audits for their tools on behalf of their clients, the ultimate responsibility for compliance lies with the employer or employment agency utilizing the AEDT.

“Although vendors of automated employment decision tools are not directly accountable under the legislation, if their clients meet the specified requirements, they may become the focus of an audit,” warns Papagianni.

Failure to undergo an audit places vendors at risk of their clients' audits revealing problematic issues, which could lead to significant fines and damage to their reputation, he adds.

Proactively seeking an audit allows software vendors to identify and address potential issues in advance, says Papagianni. “It also enables them to offer assurance to both current clients and prospective customers that their software complies with regulations,” he adds.

While HR and TA leaders should consult with their organization’s legal team, the employer is responsible, says Nicollette Nowak, vice president of legal for TA solution provider Beamery, but not entirely.

“Under NYC, local L 1 44, the employer is the ultimate responsible body,” says Nowak, “but that doesn't mean that your vendor isn't responsible.”

Nowak says employers are responsible for how they use the AI recruitment solution because they may use the tool in a way that is not intended to be used, says Nowak. “A vendor has no control over that,” she says. “You want to make sure that if you're buying an AI model from a vendor that's designed to screen people that it's not [being used incorrectly] and that model has been tested.”

“Depending who you're talking to within the organization, it's dependent on their knowledge of the current technology they use,” says a head of operations for a leading TA firm. “Depending on the stakeholder, they might know they're using ai, they might not know they're using AI.”

Who is involved in the audit? HR and TA leaders also need to be aware of what an audit involves and who plays a key role. Here's a hint: It’s more than just one person According to Bryq’s Papagianni, the individuals and parties involved in an anti-bias audit of AI-powered tools can vary depending on the organization, its specific processes, and the nature of the audit. He adds that, typically, the following parties may be involved:

Client's Legal and Compliance Teams: Responsible for ensuring audit compliance with laws, regulations, and contracts, offering legal oversight.

AI Vendor Representatives: May participate, providing technical expertise, data, and insights into AI system design.

Data Scientists and AI Experts: Analyze AI models for biases, typically from the client's organization or external consultants.

Diversity and Inclusion Specialists: Offer insights on bias, especially related to protected characteristics.

Ethics Committees or Review Boards: Oversee ethical aspects, including bias, in some organizations.

Third-Party Auditors: Independent experts specializing in AI ethics and fairness, conducting impartial audits.

Project Managers: Oversee and coordinate the audit process for timely progress.

While the initial legislation did not spell out who is considered an independent auditor, the first version of the proposed rules clarified that an independent auditor is a person or group that was not involved in using or developing the AEDT.

“However, amid concerns that this could lead to employers or employment agencies conducting internal audits of their tools, where teams not involved in the use or development of the tool would conduct the audit, the updated rules make it clear that audits should be conducted by a third party,” says Papagianni.

Once the relevant material is available, then the audit process generally takes four to eight weeks, he adds.

ChatGPT: Coming to a Candidate Assessment Test Near You

Job applicants are using generative AI for their resumes and candidate assessments. Is this cheating or the new knowledge-worker reality?

Heads of talent and recruiters have a new and modern challenge: They are not the only ones who are embracing Generative AI tools like ChatGPT - candidates are using the new breed of machine learning as well.

When ChatGPT was introduced in late 2022, the initial response centered around how it will change the working world for employers, and if they thought about the impact on employees at all, it was solely about the potential loss of jobs.

However, a growing problem is emerging for recruiters and heads of talent: Candidates are not only using ChatGPT to create resumes and cover letters, but they are also using the new advances in machine learning to ace employer assessment tests.

Assessment Vendor Bryq is not surprised. It believes that in today's job market, both job seekers and employers are constantly seeking ways to gain an edge.

“It's unrealistic to expect people not to utilize the tools at their disposal,” says Manto Papagianni, Head of Product for Bryq, a talent intelligence solution provider.

“Candidates are currently employing AI generative tools to enhance their resumes and improve their interview performance. Some also use these tools for market research, helping them explore job opportunities and identify the essential skills required for specific roles of interest,” says Papagianni.

TA leaders are taking notice. According to Robert Newry, CEO of assessment solution provider Arctic Shores, talent leaders say that they are seeing “scarily similar” applications and cover letters.

“One TA told me in one campaign they reckoned over 30% of the applications had been AI-generated,” he says.

In many ways, candidates view AI in the same light as using spell check or a calculator. “It’s super easy to use, saves you time, and complements your mental skills,” Newry adds.

But should TA leaders consider the use of AI on assessment tests as ‘cheating’? For some recruitment experts, the horse has already left the barn.

“Using tools like ChatGPT or any other technology to enhance one's profile or improve their efficiency should not be considered cheating, no more than one would consider the use of a friend’s advice or a professional resume writer as cheating,” says Bryq’s Papagianni.

TAs and HR leaders are unsure how to approach this matter. For instance, Newry says that if they allow the use of AI in the assessment process, how do they know what's human and what's AI?

“But if you ban AI, you imply you don't want to level the playing field and you are not a supporter of this game-changing technology,” he adds.

And not saying anything about the use of AI is also “dangerous,” in the words of Newry.

“Without clearly stating a position, some candidates will use AI while others won't,” he says. “You will allow a huge and hidden distortion into your candidate pool.”

Those distortions could help determine if a wave of candidates were using ChatGPT in their interviews, says Alan Bourne, Chief Scientist for Sova Assessment, a talent technology solution provider. “It would be feasible across a selection of hundreds of video interviews to check for similar answers being given that indicate a similarity to the standard Chat GPT answer to a question and where there is excessive similarity across the answer given by multiple candidates,” he says.

Claudia Nuttgens, Head of Assessment and Selection Innovation for AMS ,says HR and TA leaders are concerned about the use and abuse of AI — and rightly so, she adds.

“Some of the digital assessment tools are particularly susceptible to “cheating” with Chat GPT and it is becoming apparent that some candidate groups are really adept at using it,” she says. “We are talking about it a lot amongst ourselves as well, especially in the early careers and campus space where it is a particular issue.”

But not is all sinister about the use of AI in candidate assessments: It can often level the playing field for non-white minorities. Black and Mixed-Race students were higher users of ChatGPT in job applications than the general population average, according to Arctic Shores’ Newry.

“I've seen examples for Scrum Masters to Sales Reps and generally the higher the knowledge value in the role, the more a tool like ChatGPT can make a difference,” says Newry.

So, what can recruiters do in light of this new reality? They can overhaul their assessment tests and add a much-needed personal touch, says Bryq’s Papagianni.

“Assessments incorporating personality questions inherently resist manipulation since personality traits lack absolute right or wrong answers, and the alignment between one's personality and job prerequisites can't be directed by GenAI,” she says.

Also, a thoughtfully constructed assessment typically incorporates timed elements to accurately gauge an individual's skills. “Thereby minimizing opportunities for cheating and ensuring a reliable evaluation of their capabilities,” adds Papagianni.

Eventually, TAs could also see a return to in-person interviews and assessments.

“Clients may want to think of alternatives to the normal CV and resume approach to application for instance,” she says. “We need to think creatively about how to combat cheating but also embrace the technology and change how we communicate with clients in the process.”

Arctic Shores’ Newry agrees. He says that once TA leaders realize that ChatGPT will level the playing field, it will require a complete rethink of the hiring process.

“We need to rethink what we want from a knowledge-based role in a world where Gen AI can do a lot of the basic information processing work,” he says. “Skills like creativity, problem-solving, planning, organizing, motivating, and coordinating will be much more important than the reasoning skills we cherished in the past.”

Any resemblance?