Will GPT automate all jobs ?

And Using AI to make teaching easier & more impactful

Author

The Alignment
March 20, 2023

Happy Tuesday ! A bit of a quiet Monday but still some exciting stuff happened around the horn.

As always if you enjoy reading our posts be sure to spread the word !

Here’s what we have lined up for you today -

  1. Will GPT automate all jobs ?

  2. Using AI to make teaching easier & more impactful

  3. Sam Altman's a bit scared

  4. Scaling Verticals

Researchers at OpenAI, Open Research and University of Pennsylvania have published a paper looking at the potential labour market impact of Large Language Models.

The paper reports that irrespective of the nature of the work, about 80% of the workforce could see at least 10% of their work affected by GPTs and about 19% of the workforce could see at least 50% of their work affected by GPT.

The researchers define “exposure” to GPT as whether having access to GPT/GPT systems would reduce the time required for a human to perform a specific task by at least 50%.

The paper posits that contrary to popular belief, parts of better paying work have a higher chance of exposure to GPTs.

Furthermore the authors assert that GPTs should be classified as general purpose technologies given their potential to touch literally every part of the working economy. Hence moving forward, public policy needs to evolve alongside the evolution of these technologies for a smooth transition for the entire workforce.

The paper highlights 5 key strategies.

Strategy 1: AI-created examples

Creating a wide range of examples for concepts that are taught in class is not an easy task, AI makes this easier.

Strategy 2: AI-created explanations

A single explanation for a topic may not be understood by a diverse group of students. Coming up with lots of different ways to explain things using examples and analogies is more likely to resonate with a broader set of students. Furthermore, teachers can put in their explanations to the test with the AI to see how they can refine it.

Strategy 3: Using AI to develop low-stakes tests

GPTs are good at creating simple quizzes. Low stakes tests are a good way to assess if students are keeping up with the class. So it makes sense to create these tests with the help of AI as they can make them blazingly fast meaning that teachers can also host these at regular intervals.

Strategy 4: Assessing what students know, and what they are confused by

GPTs are great summarization tools. Teachers should take advantage of this and feed student work into GPT. Creating an overview on what the student is understanding well and what the student needs help with.

Strategy 5: Distributed practice with AI

“Distributed practice means revisiting material several times over an extended period. This approach helps students form better mental models, remember those half-forgotten facts, and really get a deeper understanding of concepts.”

Prof Mollick says that linking new ideas to concepts students already know promotes deeper learning, and AI can lend a hand by weaving past topics into lectures or discussions. By asking the AI to find relationships between concepts, teachers can present various connections between ideas, making them clearer for students.

In this wide ranging video interview about GPT-4 Sam Altman admits he is a bit scared by the possibilities of large language models. The interview is an informative watch !

Around the industry -

  1. Runway announces Gen-2, a multimodal AI system that can generate novel videos with text, images, or video clips

  2. GPT 4 jailbreaks are already happening

  3. StabilityAI releases Stable Diffusion Reimagine

Product Watch Warp AI -

AI that’s directly built into your terminal

Scaling Verticals

At this point, these curve aren't even exponentials, they're verticals !