September 14th, 2023
​

The Rebel Tech Newsletter is our safe place to critique data and tech algorithms, processes, and systems. We highlight a recent data article in the news and share resources to help you dig deeper in understand how our digital world operates. DataedX Group helps data educators, scholars and practitioners learn how to make responsible data connections. We help you source remedies and interventions based on the needs of your team or organization.

​

---

IN DATA NEWS

​

​

“Harriet Tubman, an American abolitionist and social activist, didn’t give many interviews in her lifetime, and when she did, they were generally conducted by one of her friends, Sarah Hopkins Bradford. Gillian Brockell had an "interview" with Tubman using Khan Academy’s new artificial intelligence learning tool Khanmigo. Brockell was curious to know if Khanmigo would attempt Tubman’s authentic speech, her religiosity, or her tenacity. Overall Brockell was disappointed that the bot regurgitated superficial statements and couldn't provide in-depth information about Tubman.”

​

When this article was released in July, it was met with rightful heavy criticism, especially on the ethics of using a generative AI tool on a deceased person.

The value and insights gained from these AI-generated conversations were flawed from the beginning. First, the journalist’s privileged hubris of being able to interview a legend like Mrs. Harriet Tubman still escapes me to this day. Second, as an educator, the learning objectives of this exercise are a paltry underdeveloped non-quantifiable statement: “to know if Khanmigo would attempt Tubman’s authentic speech, her religiosity, or her tenacity.” Since Mrs. Tubman didn’t give many interviews, the journalist has no correct Tubman statement to compare them against. It’s impossible for the journalist to be able to evaluate the quality AI tool’s responses. And personally, I don’t understand how a generative AI tool is useful to Khan Academy’s learning platform. Again, that’s part of her hubris, and Khan Academy surfing a trend. Third, it’s the biggie – the blatant lack of authentic words from Mrs. Tubman herself. Apparently, the Generative AI 101 lesson was missed: you need lots of quality data in order for the algorithm to give basic information.

Now that we’ve unpacked the major fallacies of this AI-generated interview exercise, let’s consider how this exercise could be improved. First, ask a living person who has ample online content e.g. written and spoken, to consent to being part of an AI-generated interview. Second, add the living person’s content to a customized and closed generative AI system. This step is very important so that the existing noisy and messy data in the systems doesn’t infect the impending conversation. Third, create a testbed of interview questions that are related to topics discussed in the content. The list of questions would increase in difficulty for the customized generative AI system to respond confidently. You’ll need a team of data professionals, instructional designers, and social scientists to make this exercise legit. When this generative AI interview happens, you’ll be able to assess the system’s responses, both qualitatively and quantitatively. And of course, the living person would be able to share their feedback on whether the AI system sounded like them or not. If you’re asked to stress test a generative AI system, follow this type of research methodology blueprint.

Read the article for yourself – no, seriously – and use it as a reference point for how not to interact with generative AI platforms.

​
​

Click Here to Read the Entire Article

Like what you're reading? Find it informative and insightful? You can sponsor the Rebel Tech Newsletter and follow on LinkedIn.

​

---

DATA CONSCIENCE CORNER

"Whether in files, spreadsheets, or the cloud, database architecture and design is leaned on as a universal fix; however, many of these decisions are being made in a vacuum." pg. 144

​
There’s a clamor for more quality data nowadays. How the data is digitally organized falls on the shoulders of data engineers, data analysts, and database administrators. And these data workers aren’t members of a company’s product/service leadership team. So data engineers, data analysts, and database administrators are making decisions about where to put each piece of content based on the *flimsy* objectives they were given with a short deadline to complete their data operations duties while dealing with a data personnel shortage. Mistakes are made. Errors are baked into the data model that transfers into the database management system and finally congeals when SQL’d into becoming a dataset. This is your friendly reminder to review and revise your database architecture and design every year. Garbage in, garbage out. Reduce the amount of data trash you accept into your systems.

​

Get Your Copy of Data Conscience Here!

​

---

​

A WORD FOR BLACK WOMEN IN DATA

Black Women in Data Summit

September 23-24, 2023

ATL | Online

$699 In-Person | $129 Online

This is the *LAST-LAST* call to book your spot to attend the largest gathering of Black women data professionals. We’re technically sold out but we’ve opened 5 more in-person spots. We know that life be life-ing. If you can now make it, come through!

​

Grab your BWD Summit tix here!

​

---

UPCOMING EVENTS

​

Elastic's Vector Search & COLLIDE Data Conference

​

If you’re in the Atlanta area, get to data learning with Elastic and Data Science Connect!

The Elastic Developer Relations Team is hosting an in-person meetup, Vector Search, presented by Alexis Roberson (Developer Advocate at Elastic) will present on Vector Search, followed by networking, light bites, and refreshments. Join them on Thursday, September 21st, 7:00-9:00 pm at Advanced Technology Development Center (75 5th St NW, Atlanta, GA). They'll be in the Community Room. Please RSVP if you plan on attending.

​

RSVP for the Vector Search here!

~

The COLLIDE Data Conference is in Atlanta, GA on Tuesday and Wednesday, October 3-4 at Center Stage Theatre in the heart of Midtown, that includes two days' worth of data-driven workshops, presentations, expert panels, a Conference Job Fair Booth, and more.

The event will feature top data science practitioners and executives from throughout the southeast and will focus on the transformation that data science is driving across industries and disciplines, from a variety of industry perspectives and levels of technical application. Get 60% off using coupon code: ‘BWD60’
​

Register for the COLLIDE Data Conference here!

​

Follow us on social

​

---

LAUGHING IS GOOD FOR THE SOUL

​

Stay Rebel Techie,

Brandeis

​

Thanks for subscribing! If you like what you read or use it as a resource, please share the newsletter signup with three friends!

​