Here is your November guide to the latest in recruitment trends, technology
ARE THE ROBOTS COMING FOR MY
A few months ago, I posed the question “Could data scientists be automated out of a job?” (check out the August edition of The Scoop). OK, a bit sensationalist, but not just for the sake of provocation — the louder and more constant conversation has focused on automation and the displacement of more task-oriented, manual labor workers and much less on the implications for the broader labor market.
However, new research out of Stanford University and the Brookings Institute takes a unique approach to this question. Looking beyond the AI tech that exists today, it assesses the overlap between the subject-noun pairs in AI patents and job descriptions to determine the level of risk exposure to automation.
From this lens, occupations composed of more high-skilled tasks actually seem to have greater AI exposure and displacement risk. In addition, gender predominance in certain occupations resulted in greater displacement risks for males vs. females, who have a greater majority in occupations with much less AI exposure.
As an example, both in support of AI’s impact on cognitive work and our distance from that reality, a team trained a neural network on a decade’s worth of Nike commercials to see if it could write an original and inspiring ad. The result, while not perfect and often laughable, had some pieces of real inspiration — I’d like to see #LegendThatThing start to trend.
The reality is that job descriptions are imperfect tools. Using tasks to give a general understanding of basic requirements doesn’t capture the evolution of companies and employees’ interests and willingness to evolve, both of which shape what a job will become.
THE POWER OF UNCONVENTIONAL THINKERS
Excited about what you’ve read on The Scoop? We’re always looking for people who are just as interested in these topics as we are! Join TMP and help us shape the future of recruitment marketing.
ARE MORE THAN 7 MILLION PEOPLE MISSING YOUR MESSAGE?
There are roughly 7 million people with a visual disability in the United States, and over half of working age people who are blind or visually impaired are unemployed.
A review of websites, ranging from ecommerce to news to government services, revealed that about 70% contained “accessibility blocks,” or quirks in the design that make them unreadable with assistive technology.
In an independent third-party study of career site accessibility — an important consideration in an inclusive candidate experience — TalentBrew outperformed the six largest recruitment marketing platforms by an average of 23%. Click To Tweet
Some tech companies are making aggressive efforts toward greater web accessibility, making platform improvement updates based on suggestions from the Blind Accessibility Testers Society (BATS).
The downstream effect of inclusive design efforts is better SEO and a structural foundation that will power future AI-based innovations.
Google even has a centralized accessibility team that works to codify inclusion into every product. And they’ve recently partnered with the Canadian Down Syndrome Society to better train the Google voice assistant to understand users with Down syndrome.
Interested in learning more about web accessibility and the potential impact to job seekers? Drop us a line at firstname.lastname@example.org. TMP’s Director of Accessibility, Mike Spellacy (lovingly called Spell), is always excited to share his expertise on this topic.
WHEN IT COMES TO ALGORITHMIC BIAS, IS LESS ACTUALLY … LESS ACCURATE?
Back in May, I wrote about a move by the College Board to include data aligned with hardship into the admissions process as a way to give context to SAT scores and college readiness (check out the May edition of The Scoop, Factoring Adversity into Selection). Recently, the Apple Card has come under scrutiny for what appears to be gender discriminatory credit decisions — which calls into question the practice of hiding (often called “blind”) versus reintroducing characteristics like gender and race into the model as a way of counteracting bias.
One research study created two separate analyses using machine learning models to predict creditworthiness. 93% of women got more credit using a model that was trained on only women’s data versus one where men and women were mixed together. Women were actually more likely to pay back their loan, but that factor indicating greater credit worthiness was not accounted for in the combined model based on the averages of men and women.
There are challenges with this approach as well, because it opens up an opportunity for intentional discriminatory behavior (e.g., Facebook’s challenge with discriminatory housing ads), but these are the conversations that will move us past using technology for blinding characteristics as a means of accounting for unconscious bias, toward greater visibility and personal accountability for those decisions.
ROUNDING OUT THE SCOOP: PSYCH, SOCIAL, LABOR
- Instagram rolls out a TikTok function in Brazil, Instagram Reels. Will it come to the US soon? https://techcrunch.com/2019/11/12/instagram-reels/
- Holograms you can see, hear… and feel! New technological advancements may change the way we think about engagement. https://futurism.com/scientists-create-holograms-see-hear-feel
- Could creating Hunger Games-like conditions for robots be the answer to performance improvement? https://www.popularmechanics.com/technology/robots/a29831507/robots-mortality/