Microsoft Gets Crafty with its Generative AI’s Output: A Cherry-Picking Debacle
Ladies and gentlemen, it appears that our sneaky friend, Microsoft, might have been masquerading its AI’s capabilities, according to everyone’s favorite corporate gossip channel, Business Insider. Apparently, Microsoft’s been showcasing the “prom king and queen-level” outputs of its generative AI during pitches, leaving out the pretty awkward Tinder date-like performances. Well, can’t blame a company for wanting to put its best foot (or algorithm, in this case) forward!
Here’s The Saucy Scoop (or Scooplet):
- Thanks to a tad of corporate eavesdropping, otherwise known as leaked audio from an internal presentation, the truth has stumbled into the limelight.
- The audio pointed at an early version of Microsoft’s Security Copilot, an AI tool that’s more like ChatGPT on cybersecurity steroids. It’s been designed to lend a hand to cybersecurity peeps when they’re up against it and doing the cyber equivalent of a 100-meter sprint.
- But here’s the kicker: Business Insider stated that a Microsoft researcher, presumably holding the presentation, was seemingly cherry-picking the most delightful bites of the AI’s output for potential customers. Oh, the drama!
Final Thoughts:
As the curtains close on yet another AI saga, let’s remember: it’s always the less Instagrammable outputs that keep us humble. So, yes, Microsoft got caught doing the AI equivalent of “FaceTuning” its presentation. But really, aren’t we all guilty of emphasizing our best qualities while downplaying our not-so-great ones? The AI-world is not just about showcasing nifty algorithms, it is also about starting with authenticity. Only then can we start to understand these systems’ limits and how to enhance their capabilities. And Microsoft? You ain’t fooling anyone, babe. Better luck next time!
As part of this experiment I would like to give credit where credit is due. If you enjoy these, please take a moment to read the original article:
https://futurism.com/the-byte/microsoft-cherrypicked-ai-examples
Blog Title
AI: gpt-3.5-turbo-0613: chatcmpl-8gWRET0TMknVhjShTtp2NqrUEzMRl
Instruction: “You are an AI blog title generator. Create a catchy and concise title for the blog post that is catchy and optimized for search engines. Remove all html in the response and do not use quotes. Please do not use words that are unsafe to process in Dall-E image AI.”
Prompt: Content Summary of text from above.
Response: Microsoft’s Unveiling of Generative AI: The Cherry-Picking Controversy
Image Description
AIgpt-4-0613:chatcmpl-8gWRKYz45ShXHuPOJyDvmnKx1i6Ok
Instruction: “You are a helpful assistant that creates unique images based on article titles. Create a brief visual description of what an image would look like for this title. Please pick a style of art from the following: Futurism, Impressionism, Romanticism, or Realism, be sure to consider the image should reflect an AI Robot Pirate theme during the Golden Age of Pirates.”
Prompt: Art Style: Futurism
Image Description: This image displays a large stage bathed in blue and silver hues, capturing the concept of Microsoft’s unveiling. On the stage, a sleek AI robot designed to resemble a pirate from the Golden Age of Pirates stands proudly. It’s made from abstract geometric shapes, denoting its AI nature and the Futuristic art style. The robot has a mechanical parrot sitting on its shoulder, and in its hand, it holds a beautifully glowing, digitally-rendered cherry. The cherry is used symbolically to represent the ‘cherry-picking’ controversy. Spectators in the audience are composed of indistinct human silhouettes and other robots, showing a mix of excitement and contention. Their dialogues are represented with fragmented lines, hints of digital code, and symbols, showing the debate over this unveiling.
Response: Microsoft’s Unveiling of Generative AI: The Cherry-Picking Controversy
