Alexandra Samuel, Ph.D. on LinkedIn: The missing feature in every writing and editing tool - Josh Bernoff (2024)

There was a time when we didn't all live in fear of having our worst moments captured on #socialmedia and shared with the world. I'm honored to talk about those more innocent days in Asia Youngman and Kathleen Jayme's brilliant film, "I'm Just Here for The Riot", now airing on TSN - The Sports Network - Canada and ESPN. Their movie is a must-watch for #Vancouver folks, #hockey fans and people who are interested in the evolution of social media and digital culture. It looks at the 2011 Stanley Cup riot in Vancouver, which was the first time a public disturbance was captured on mobile phones and then posted to social media in the effort to bring the rioters to "justice".I put "justice" in quotes because it was immediately apparent to me that this was a dangerous use of social media, and a step towards the kind of citizen-on-citizen surveillance that is a hallmark of authoritarianism. I wrote a piece to that effect for the Harvard Business Review, the day after the riot, and for a couple of weeks I took a lot of flack for criticizing what many people saw as an empowering grassroots movement for justice.Within a couple of weeks it became clear that naming and shaming on social media was the opposite of justice. People lost jobs and safety because photos of them intervening in an effort to stop people from engaging in criminal activity were misconstrued as evidence that they'd been rioting themselves. And as you can see in the documentary, a lot of young people who did participate in the riot—because teens can be stupid, and caught up in the moment, just like any of us—had their lives derailed in a way that was perhaps disproportionate to their crimes."I'm Just Here for the Riot" does a terrific job of capturing the tragedy and complexity of that moment, and how it set the stage for the way social media has been used in many subsequent incidents of shaming and vigilante justice. You can watch the movie on TSN this weekend, or watch it via ESPN or TSN on demand. Congratulaions Asia and Kat for an amazing film!

3

1 Comment

Do you have a “work spouse”, “work wife” or “work husband”? And if so, have you ever had to navigate conflict in your work marriage—or worked to avoid it?I would love to hear your thoughts or stories for an article I am developing. I won’t quote you without permission. Feel free to message me, email me or leave a comment.

7

It is time to stop treating #AI like a gumball machine: Prompt in, magic out.That insight was part of my wide-ranging conversation with Daan van Rossum of FlexOS, looking at how we can start approaching AI at a team level.Watch, read or listen here: https://lnkd.in/g56D_EYg

17

6 Comments

Nonprofit friends, if you can take a moment to answer this survey you can help make future surveys better!

1 Comment

What if you could make your #customerservice interactions INSTANTLY easier, with the help of #AI? My latest for The Wall Street Journal shares my vision for making that happen, with a chatbot that represents customers just the way so many chatbots now represent brands.This piece was inspired by an interaction I had online a couple of months ago, when I swung into action after my mom was misled into signing up for an online service that charged her much more than promised. After a little googling I saw that many (mostly older) people reported similar experiences with this service; those who relentlessly pestered the customer support line for a refund eventually prevailed.As I joined the chat line and started the pestering process myself, I thought: Why should I sit here, pestering? Why not write a script to do the pestering for me? So I fired up #ChatGPT and started writing a script.Then it occurred to me to let the company support agent know what I was working on. "I can see from Reddit that your strategy is to just avoid the refund as long as possible while I pester and pester," I wrote. "I am writing a script to automate the pestering process here and I am going to post it online so other people can pester you when you mislead them. I would rather not spend my time on this but if i have to sit here waiting for you, I might as well write a pester script and share it with others."Surprise, surprise: No further pestering was required. The support agent INSTANTLY refunded my mom.That made me think: What would happen if this became the new norm in support interactions? What if we, the customers, used automation just as powerfully as the companies that would chatbot us into exasperated defeat?A pesterbot is just one way automation could make our lives as customers much easier. My article shares the rest of my wishlist. (And it has the BEST illustration!) Link in comments.

24

13 Comments

When it comes to #AI, beware of hallucinations. Not just AI hallucinations: human hallucinations! As someone who writes about AI for a newspaper, here's how I spot writers (and workers) who are hallucinating in their take on AI.Hallucination #1: Wishful thinking.If you want to believe that AI is not capable of replacing you or your work, you will definitely be able to engineer your AI interactions to confirm that hypothesis. Instead, try to prove yourself wrong by trying many experiments at self-replacement. That's more likely to yield an accurate result.Hallucination #2. Time dilation.Expecting AI to deliver great results instantly is another delusion. Humans are what slow the process down: AIs may work in nanoseconds but it takes us hours, months or days to learn how to coax great results from the machine, or to figure out where these new technologies are useful. Reserve a verdict until you have tried multiple technologies over many months with many use cases/approaches.Hallucination #3: Universalizing.Don't imagine that your personal experience with AI is a broadly relevant indicator of AI's ultimate value or utility. Instead of universalizing from your personal experience, get curious about how other people are using it, and why they are finding it valuable.What hallucinations have you witnessed among human AI users? I'd love to hear what you observe in the eager dismissals of AI.

9

3 Comments

Stop doing the tedious tasks that #AI can do for you. My cover story for The Wall Street Journal's report on Artificial Intelligence shared the why: Today, I'm sharing the how—and a link to the custom GPT that will do this for you.Now that custom GPTs are available to anyone (even if you're not paying for an #OpenAI subscription), anyone can use custom GPTs like the one I describe in my story. It's a very simple #GPT that converts a supplier's free-form email invoices into a structured table that I can paste into Excel and load into my #bookkeeping app. It's incredibly easy to set up yourself: My latest newsletter (link in comments) has a screenshot with the complete instructions, including the "before" and "after" that tell the Invoicerizer what its results should look like. The core instructions are..._______Take free-form email invoices and convert to a table format for financial recordkeeping. The table should include the following columns:Date | Hours on site | Hours for day | Cumulative hours | Task description | Cost for the Day | Cumulative TotalIf there is a day marked "off," mark "off" under "hours on site". Use the before and after examples to determine how to format and include the daily and cumulative billing amounts calculated at $[rate]/hr.[Then provide "before" and "after" examples.]____I'll post a link to my custom GPT in comments.

43

3 Comments

This is the kind of garbage operation that makes people so suspicious of where #AI will take us. The heart of the matter: “Basically, AdVon engages in what Google calls "site reputation abuse": it strikes deals with publishers in which it provides huge numbers of extremely low-quality product reviews — often for surprisingly prominent publications — intended to pull in traffic from people Googling things like "best ab roller." The idea seems to be that these visitors will be fooled into thinking the recommendations were made by the publication's actual journalists and click one of the articles' affiliate links, kicking back a little money if they make a purchase.”link in comments.

8

1 Comment

How can #AI help you work more effectively with #spreadsheet, #Excel or #CSV files? A few readers have asked for details based on my story for The Wall Street Journal this week, so here are some practical tips and examples.Sign up for my newsletter ☝ to get more details on my stories for the WSJ and for lots of AI how-tos. I'll put a link to the latest story in comments. Here's the scoop on how to do the kind of thing I wrote about this week.How to get spreadsheet data into an AI-friendly form:✅ Upload CSVsto ChatGPT orClaude.aias part of a prompt or custom GPT📊 Paste data from Excel sheet (or uploaded a csv intoCoda)⬆️ Upload an xlsx file to Google Sheets use a GPT add-on💻 Try an app designed for working with spreadsheet data (like Akkio or Numerous.ai, though so far I haven't found them useful enough to be worth paying for)💵 Pro tip: Conserve tokens when asking GPT to work with a CSV by adding an index column to your input file (e.g. IN1, IN2, IN3). Then ask for it to return results (like categorization of your original data) as a 2-column table in the form Index No | Category. That way you don't use up tokens/answer power just returning your original data. Use XLOOKUP to re-unite the categories with your original dataset. (ChatGPT can tell you how to do that.)How I've used AI to work with spreadsheet data📁 Categorizesocial media postsby topic (small sets viaCoda.io, larger via CSV upload to GPT and then asking for results in table form).🧹 Clean a messy export of past articles by returning a file as a de-duped table.⚾️ Power a custom GPT that helps me draft story pitches. I uploaded a CSV of social media analytics for all my past published stories so the AI knows what performs best.🎓 Digest the stats in an academic article by returning the table in a PDF in CSV form that's easier to read and work with.🔎 Make my ChatGPT history searchable. I exported, got GPT's help writing a Python script to convert to CSV, uploaded to Coda.io, and used Coda's on-board AI to categorize past chats. 🐍 Learned how to use Python to clean up an Excel file! I had a file with 8 tabs of similar but differently labeled data, so I tookall the separate column headings and pasted them into a single CSV. Then I pasted that into GPT, along withthe list of what I wanted as my canonical column headings, and asked it to map each tabs column headings onto my canonical headings and return as a table. It got that about 90% correct, but I had to manually correct the mappings. But THEN my mappings were the only input I needed in orderto write a Python script that took my many tabs and consolidate them into a single sheet withconsistent headings! And I did all that as someone who basically does not know Python; it was just monkey see, monkey do, withme following GPT'sinstructions. Are you using AI to work with spreadsheets? I'd love to hear how.

5

1 Comment