3 Min Read • Intensity: ☕️ 

Happy Monday! Caffeinators☀️

As you sip your coffee, we have 3 bad news and 1 good news waiting for ya! (Well… depends on how you see them)

Today’s Starting Sips:

• 🎭 AI Impacting Politics… Again

• 🐶 The “Sniffer-Dog” of GPT

• 📉 Is GPT Dumbing Down?

🎭 AI Impacting Politics… Again

AI technology has found its way into presidential politics again. On June 8th, Ron DeSantis released a campaign video containing images of Donald Trump and Anthony Fauci that suggested a close bond. However, not all of the pictures are genuine.

Take a look at some of the fake ones:

Under closer inspection, there are signs of AI manipulation — Trump’s unusually smooth hair and the blurry backdrop. It's so subtle that most fall victim to the “artful” deception…

These crafty AI-generated images underscore a new age threat: AI-facilitated disinformation. At CaffeinatedGPT, we want to raise awareness for AI threats as we sail into uncharted territory. The road ahead is a blur of reality and AI fabrication.

Want to see how easy it is to create misinformation? Look what I made in only 2 minutes:

🐶 The “Sniffer-Dog” of GPT

Beware, student! Kansas University professors have coded an AI "sniffer-dog" for papers penned by GPT, claiming a near-perfect 99% accuracy!

But don't shred your GPT drafts just yet! This algorithm gets thrown off by papers containing human edits, so you just have to let GPT do 98% of your paper now 🤣

Plus, its test samples were limited to a mere 180 papers, and accuracy dipped to 92% on paragraph-level checks.

Though not the full AI-buster package yet, It’s still an impressive retaliation! The road to AI authentication may be bumpy, but we're surprisingly well on our way.

Perhaps soon, we'll even have a tool to spot AI-crafted images.

Remember Professor Mumm and his GPT witch-hunt? Bet he already put this tool on his wish list 🎅 

📉 Is GPT Dumbing Down?

Users across OpenAI's community are questioning whether the latest update left our AI friend with fewer IQ points. Almost like GPT’s been missing its morning coffee lately.

OpenAI Community user @radiator57 presents a compelling case: GPT seems to have developed a conversational memory of a goldfish. It once flawlessly held a conversation over 30 exchanges, but now loses its memory after as little as 15 messages.

It gets worse: logic errors are on the rise, and language consistency is down the drain. To make matters worse, GPT even has conversational mood swings, suddenly jumping from Shakespearean to street lingo.

But the big question remains: what's causing this? Cost-saving measures at OpenAI? New preventive measures limiting GPT's intellectual freedom? Your guess is as good as ours. Only time (and hopefully more updates) will tell!

However, let's be clear: these concerns are mainly anecdotal. OpenAI hasn't officially addressed these observations yet. Until then, let's hope our beloved AI chat bud gets its groove back soon!

So, what's the plan while we wait? I've put together 4 excellent prompt strategies designed to significantly boost your productivity! These will be released tomorrow at the same time! So, get ready 😄

🍿 Video of the Day

It’s Monday so we picked a 3min comedy to help you de-stress 😄

🥤 To-Go Cup

• AI has once again infiltrated politics, this time through misleading images involving Trump and Dr. Fauci

• Professor at Kansas University developed a prototype of an AI 'sniffer-dog', capable of identifying research papers written by GPT

• Members of the OpenAI community have consistently reported a reduction in performance for both ChatGPT and GPT-4

If you enjoyed this article, I think you will like this one too! It’s about Sam Altman warning the US Senate about AI threats.

Until next time,

Your friendly barista and ☕️ GPT!