I admit it.

I get irritated by those emails, documents and presentations that demonstrate very little, or no, critical thinking on behalf of the “author”.

In a busy workday, we want to feel confident that the important information and opinions provided are those of, or mediated by, the brains of your colleagues across the business.

That niggling concern that the material you received has not been fully validated, and may result in unexpected/unrequested workload transfer and/or duplication of effort. 

Worst case, in a report I read this week, is that your customers may become the “beneficiaries” of time-wasting and misleading communications.

The article, by Joe Procopio, reported the news of the first company-wide AI (LLM) ban.

Overreaction maybe, but it reflects growing concerns.

“The whole sordid story started when a linguistically obtuse and factually inaccurate AI-written email that was purported to be personally written by a customer-facing human was whistle blown to the CEO by one of his customer support agents, who got it from an irritated customer.”

Digging into this specific example;

• The company neither mandated nor restricted AI agent use, and while they weren’t running amok, they weren’t being managed properly either. In this case, there was no executive mandate to use AI and so there was no “walking back” the game-changing benefits. Nothing holding them back from a “scorched earth” policy.  

• The company reportedly instituted a total ban, a complete cease-and-desist, do not engage with LLMs at any time, for any reason, across the entire organization.

Genius or tantrum?

Debapratim De reported, in a side note to his Deloitte Monday Briefing, an interesting, related statistic and resultant effect.

• Average monthly research paper submissions to arXiv (a popular repository for scientific papers) hit a record high in May, and is now almost 67% higher than in November 2022, when ChatGPT was launched.
• A study shows that the use of LLMs has driven a sharp rise in submissions of manuscripts, especially from junior researchers. These drafts can be linguistically complex but are often of poor quality.
• As a result, arXiv now requires first-time authors to seek endorsement from an established academic before accepting their drafts and instituted a one-year ban on authors who submit papers with unverified AI-generated content (including hallucinated references and accidentally included user prompts 😉).  

Te-Ping Chen in The Wall Street Journal reports, “Writers Are Going to Extremes to Prove They Didn’t Use AI”

• As AI-generated writing floods our workday (and beyond), greater discernment means that there is a growing focus on detecting AI generated narrative.
• This is resulting in a growth, even in those who acknowledge using AI for help, of attempting to master something they never worried about before, “how to sound human”.
• This sounds like a “Reverse Turing Test”. 
  • Flipping the rules of Alan Turing’s classic experiment. Instead of a machine trying to prove it is human, a human tries to prove that they are human . . . . 😉
• A growing number of professionals fear being accused of wielding “machine-made material” and are proactively trying to prove their human bona fides.
• “It’s like the new McCarthyism, it’s just crazy”. “People are demanding proof of something that can’t be proven.”

It is worth thinking through these examples and contrasting to our own experiences.

How and where do we optimise AI usage to unleash genuine enterprise value?

• There is a governance and guidance issue – what business outcome are we aiming for?
• There is a question of applicability – when should we use which tools? 
• There is a critical thinking issue – how do we avoid atrophy?
• There is an expectations issue – are we sending the right message?

“As AI costs continue to be unmasked, AI productivity continues to be called out, and AI-as-labor-replacement continues to look like a limited option, there is a growing shift from “hype circus” to business-value focused implementation”.

That is good news, but what is the answer to the issues raised?

Not a ban for sure.

But clarity is required,  expectations on mindsets and critical thinking, how we collaborate, treat everyone’s time as valuable. Don’t push your incomplete tasks to your colleagues and customers – maybe this touches on the Customer Experience (CX) debate also?

It’s not rocket science. It’s just courtesy.

You can read Joe Procopio’s article “The First Company-Wide AI Ban Just Hit My Inbox. Here’s What It Means” here . . . 

You can read Te-Ping Chen’s article in WSJ “Writers Are Going to Extremes to Prove They Didn’t Use AI“ here . . . 

Debapratim De’s “Monday Briefing” on the web does not include the charts and statistics I quoted above from his email sadly, but you can access to his weekly note here  . . . . 

Thanks for reading . . . .