AI Disrupts Edtech Giants

Plus: Meta discovers AI vulnerability. Midjourney 5.1 is out.

Hello, Engineering Leaders and AI enthusiasts,

Welcome to the 11th edition of The AI Edge newsletter. Today, we bring you AI’s role in the downfall of Edtech. Thank you everyone who is reading this.

In today’s edition:

📉 AI disrupts Edtech space triggering stock falls
🔍 Meta researchers find AI “Déjà Vu”ing
🚀 Midjourney 5.1 released
🤖 AI Agent Basics: Let’s Think Step By Step by

Jon Stokes

Let’s go

---

AI disrupts Edtech space: Chegg, Pearson, Duolingo, and more saw a steep fall in stock prices

Stock prices of Chegg, Pearson, and Duolingo plummeted by nearly 50%, 15%, and 10%, respectively. The downfall was triggered Monday evening after Chegg CEO Dan Rosenweig’s comment on AI’s impact on Chegg’s new customer growth rate

Ever since AI’s arrival, there has been a continuous discussion on using AI for education. Sal Khan, founder and CEO of Khan Academy, said, “We're at the cusp of using AI for probably the biggest positive transformation that education has ever seen." He also mentions the use of personal AI tutors for every student in his Ted Talk.

Why does this matter?

Edtech could very well be the worst-affected industry as AI takes over. Right now, the only thing working in favor of edtech companies is the inaccuracy of AI models. A lot of AI chatbots confidently provide wrong answers and hallucinate. However, as AI models improve and address the issues, the situation could worsen for edtech business owners.

Source

---

Meta researchers find AI “Déjà Vu”ing: Suggested ways to address the privacy risks

Researchers at Meta recently discovered an anomaly common across most Self Supervised Learning (SSL) algorithms and call it Déjà Vu. They said SSL models can unintendedly memorize specific parts in individual training samples rather than learning semantically meaningful associations.

The report shares the details of studies around this unintended memorization and also explores ways of avoiding it.

Why does this matter?

AI models are fed huge volumes of data. The data is usually meant to train the model and definitely not to share with others, especially when it belongs to businesses and individuals. Unintended memorization could potentially leak training data. However, it’s positive news that the issue has been detected already.

Source

---

Midjourney 5.1 Release Brings Exciting Advancements to AI Art

Midjourney has released version 5.1, which brings significant improvements in the quality of results. The noteworthy enhancements are: 

• Enhanced coherence

• Improved accuracy in response to text prompts

• Reduced unwanted borders and text artifacts

• Sharper image quality

Why does this matter?

Midjourney has already captured a huge chunk of the market when it comes to AI-powered image generation. It seems the recent improvement will only solidify its position. This goes on to show the massive opportunities that come with AI use cases.

Source

---

Knowledge Nugget: AI Agent Basics: Let’s Think Step By Step

In this article,

Jon Stokes

breaks down the core concepts behind autonomous AI agents. He dives deep into the two reasoning ability of LLMs,

1. Chain-of-thought reasoning, where the model can be prompted to walk through the logic behind some conclusion or output. (This is the famous “let’s think step-by-step” prompt trick.)

2. Action planning reasoning, where the model can be prompted to come up with a series of steps that will lead to some future goal.

He then uses the two to explain how an AI agent can make decisions autonomously. It’s an extremely interesting read and makes the concept of AI agents a little less mind-boggling.

Why does this matter?

AI agents might become our go-to way to complete tasks. They are extremely impressive even at their infancy and will only get better with time. If you understand their underlying framework and capabilities, you’ll know what nature of tasks are more suited to these AI agents.

Source

---

What Else Is Happening

😲 NVIDIA’s Graphic Research impresses everyone. (Link)

🧠 Solution to eliminate hallucinations in LLM. (Link)

🤖 New Relic launches its AI observability assistant, Grok. (Link)

🚀 Mojo claims to be the newer, faster AI language for AI developers. (Link)

🏥 AI can detect diseases in real-time. (Link)

🏠 ChatGPT plugin that finds you a house. (Link)

📊 92% of businesses use AI for personalization. (Link)

---

Trending Tools

• Hack social: Gauge social media influence on Twitter & LinkedIn with Clout Check. Track & boost digital impact.

• Simbla: Create a tailored CRM with Simbla’s revolutionary AI platform. No more long implementation projects.

• Pixelfy: Create breathtaking pixel art with Pixelfy’s open-source AI-powered tool. Featuring a cutting edge prompt builder.

• Repromptify: No need for prompt engineers. Tell this AI what you want and get perfect prompts for ChatGPT, DALLE•2, and Midjourney.

• Emblify: Bring your logo to life in new and exciting settings with Emblify AI. Elevate your brand’s social media presence.

• Parallel AI: Work smarter with custom-built features. Build virtual companies with AI employees and subject matter experts.

• Value Proposition Canvas: AI tool for Value prop canvas. Input business description and get canvas outlining customer’s jobs, pains, and gains.

• Talkpal: GPT-powered AI language tutor. Chat about unlimited topics with realistic voice messages.

---

That's all for now!

If you are new to ‘The AI Edge’ newsletter. Subscribe to receive the ‘Ultimate AI tools and ChatGPT Prompt guide’ specifically designed for Engineering Leaders and AI enthusiasts.

Thanks for reading, and see you tomorrow.

---