Technology
How Smart is ChatGPT?
Visualizing ChatGPT’s Performance in Human Exams
ChatGPT, a language model developed by OpenAI, has become incredibly popular over the past year due to its ability to generate human-like responses in a wide range of circumstances.
In fact, ChatGPT has become so competent, that students are now using it to help them with their homework. This has prompted several U.S. school districts to block devices from accessing the model while on their networks.
So, how smart is ChatGPT?
In a technical report released on March 27, 2023, OpenAI provided a comprehensive brief on its most recent model, known as GPT-4. Included in this report were a set of exam results, which we’ve visualized in the graphic above.
GPT-4 vs. GPT-3.5
To benchmark the capabilities of ChatGPT, OpenAI simulated test runs of various professional and academic exams. This includes SATs, the bar examination, and various advanced placement (AP) finals.
Performance was measured in percentiles, which were based on the most recently available score distributions for test takers of each exam type.
Percentile scoring is a way of ranking one’s performance relative to the performance of others. For instance, if you placed in the 60th percentile on a test, this means that you scored higher than 60% of test-takers.
The following table lists the results that we visualized in the graphic.
Category | Exam | GPT-4 Percentile | GPT-3.5 Percentile |
---|---|---|---|
Law | Uniform Bar Exam | 90 | 10 |
Law | LSAT | 88 | 40 |
SAT | Evidence-based Reading & Writing | 93 | 87 |
SAT | Math | 89 | 70 |
Graduate Record Examination (GRE) | Quantitative | 80 | 25 |
Graduate Record Examination (GRE) | Verbal | 99 | 63 |
Graduate Record Examination (GRE) | Writing | 54 | 54 |
Advanced Placement (AP) | Biology | 85 | 62 |
Advanced Placement (AP) | Calculus | 43 | 0 |
Advanced Placement (AP) | Chemistry | 71 | 22 |
Advanced Placement (AP) | Physics 2 | 66 | 30 |
Advanced Placement (AP) | Psychology | 83 | 83 |
Advanced Placement (AP) | Statistics | 85 | 40 |
Advanced Placement (AP) | English Language | 14 | 14 |
Advanced Placement (AP) | English Literature | 8 | 8 |
Competitive Programming | Codeforces Rating | <5 | <5 |
The scores reported above are for GPT-4 with visual inputs enabled. Please see OpenAI’s technical report for more comprehensive results.
As we can see, GPT-4 (released in March 2023) is much more capable than GPT-3.5 (released March 2022) in the majority of these exams. It was, however, unable to improve in AP English and in competitive programming.
Regarding AP English (and other exams where written responses were required), ChatGPT’s submissions were graded by “1-2 qualified third-party contractors with relevant work experience grading those essays”. While ChatGPT is certainly capable of producing adequate essays, it may have struggled to comprehend the exam’s prompts.
For competitive programming, GPT attempted 10 Codeforces contests 100 times each. Codeforces hosts competitive programming contests where participants must solve complex problems. GPT-4’s average Codeforces rating is 392 (below the 5th percentile), while its highest on a single contest was around 1,300. Referencing the Codeforces ratings page, the top-scoring user is jiangly from China with a rating of 3,841.
What’s Changed With GPT-4?
Here are some areas where GPT-4 has improved the user experience over GPT-3.5.
Internet Access and Plugins
A limiting factor with GPT-3.5 was that it didn’t have access to the internet and was only trained on data up to June 2021.
With GPT-4, users will have access to various plugins that empower ChatGPT to access the internet, provide more up to date responses, and complete a wider range of tasks. This includes third-party plugins from services such as Expedia which will enable ChatGPT to book an entire vacation for you.
Visual Inputs
While GPT-3.5 could only accept text inputs, GPT-4 has the ability to also analyze images. Users will be able to ask ChatGPT to describe a photo, analyze a chart, or even explain a meme.
Greater Context Length
Lastly, GPT-4 is able to handle much larger amounts of text and keep conversations going for longer. For reference, GPT-3.5 had a max request value of 4,096 tokens, which is equivalent to roughly 3,000 words. GPT-4 has two variants, one with 8,192 tokens (6,000 words) and another with 32,768 tokens (24,000 words).
Interested in learning more about the impact artificial intelligence is having on the world of work? VC+ members have access to this special dispatch as well as our entire archive of VC+ content. Find out more. |
Technology
Charted: The Jobs Most Impacted by AI
We visualized the results of an analysis by the World Economic Forum, which uncovered the jobs most impacted by AI.
Charted: The Jobs Most Impacted by AI
This was originally posted on our Voronoi app. Download the app for free on iOS or Android and discover incredible data-driven charts from a variety of trusted sources.
Large language models (LLMs) and other generative AI tools haven’t been around for very long, but they’re expected to have far-reaching impacts on the way people do their jobs. With this in mind, researchers have already begun studying the potential impacts of this transformative technology.
In this graphic, we’ve visualized the results of a World Economic Forum report, which estimated how different job departments will be exposed to AI disruption.
Data and Methodology
To identify the job departments most impacted by AI, researchers assessed over 19,000 occupational tasks (e.g. reading documents) to determine if they relied on language. If a task was deemed language-based, it was then determined how much human involvement was needed to complete that task.
With this analysis, researchers were then able to estimate how AI would impact different occupational groups.
Department | Large impact (%) | Small impact (%) | No impact (%) |
---|---|---|---|
IT | 73 | 26 | 1 |
Finance | 70 | 21 | 9 |
Customer Sales | 67 | 16 | 17 |
Operations | 65 | 18 | 17 |
HR | 57 | 41 | 2 |
Marketing | 56 | 41 | 3 |
Legal | 46 | 50 | 4 |
Supply Chain | 43 | 18 | 39 |
In our graphic, large impact refers to tasks that will be fully automated or significantly altered by AI technologies. Small impact refers to tasks that have a lesser potential for disruption.
Where AI will make the biggest impact
Jobs in information technology (IT) and finance have the highest share of tasks expected to be largely impacted by AI.
Within IT, tasks that are expected to be automated include software quality assurance and customer support. On the finance side, researchers believe that AI could be significantly useful for bookkeeping, accounting, and auditing.
Still interested in AI? Check out this graphic which ranked the most commonly used AI tools in 2023.
-
Money7 days ago
Visualizing All of the U.S. Currency in Circulation
-
Stocks2 weeks ago
Ranked: South Korea’s Largest Companies by Market Capitalization
-
VC+2 weeks ago
What’s New on VC+ in March?
-
Markets2 weeks ago
Confidence in the Global Economy, by Country
-
Wealth2 weeks ago
Mapped: Where Do the Wealthiest People in the World Live?
-
Misc1 week ago
Ranked: Global Airlines with the Most Plane Crashes
-
Technology1 week ago
Visualizing iPhone 15 Production by Manufacturer in 2023
-
Automotive1 week ago
Visualizing Global Electric Vehicle Sales in 2023, by Market Share