Industry Insights

Background image of user typing on a calculator with floating interface elements surrounding them

January Monthly News Digest: Data Science & Analytics

Posted
January 30, 2023

Welcome to Burtch Works' Monthly News Digest, your go-to source for a comprehensive overview of the month's most significant stories in the world of data science and analytics. This month, we delve into the expanding impact of AI on various fields and industries, from healthcare and education to business and law. With our digest, you'll get a quick yet comprehensive breakdown of the latest news in the field, including the capabilities and potential uses of language models like ChatGPT, the ethical considerations surrounding AI, and other key developments in the field of data science and analytics. Whether you're a data scientist, an analytics professional, or simply someone who is curious about the latest advancements in this exciting field, this digest is the perfect resource for staying informed on the most important stories in data science and analytics.

ChatGPT Is Passing the Tests Required for Medical Licenses and Business Degrees

ChatGPT has caused concern among teachers and academics as it has passed exams required for obtaining medical licenses, business degrees, and even components of the bar exam with a high degree of accuracy. A study conducted by a Wharton Professor used OpenAI's GPT-3, the language model on which ChatGPT is built, to take a final exam of a core MBA course, and the model would have received a B to B- on the exam.This has led to a conversation about the potential use of ChatGPT as a resource in education and healthcare, rather than a replacement for certain skills. The researchers have also concluded that large language models like ChatGPT have the potential to assist with medical education and decision-making. Clinicians within a virtual clinic called Ansible Health have already begun experimenting with ChatGPT in assisting with writing tasks, such as composing appeal letters and simplifying complicated medical reports to help patients better understand their conditions. ChatGPT's ability to take high-level exams without years of studying and obtaining degrees is beginning to destabilize the measurements used to determine academic success.Read the full article here.

20 Questions (with Answers) to Detect Fake Data Scientists: ChatGPT Edition

Seven years ago, KDnuggets published an article titled "20 Questions to Detect Fake Data Scientists" which aimed to distinguish between real and fake data scientists. On the seventh anniversary of the original article, KDnuggets decided to resurrect those same questions and posed them to ChatGPT, a language model developed by OpenAI. The article notes that ChatGPT can be used to generate responses to these questions, but cautions that the model's answers should be taken with a grain of salt. Answers to questions like these should not be considered definitive proof of a person's expertise.The questions cover a wide range of topics, including data regulation, predictive modeling, precision and recall, algorithmic improvement, and root cause analysis. Despite the questions being mostly definitional in nature, the article highlights that ChatGPT does a great job of discovering and conveying well-written answers. Questions that require an understanding of the topics behind the definitions might be more difficult for ChatGPT to produce a valid answer for.Read the full article here.

Top AI Conference Bans ChatGPT in Paper Submissions (and Why it Matters)

Well, that didn’t take long! Shortly after ChatGPT was released, people started pushing the envelope of what it can be used for immediately. Already, there is some backlash, which led to a machine learning conference explicitly banning ChatGPT’s use for paper submissions. Basically, a machine learning conference has banned some uses of machine learning. Quite a situation indeed.Underneath the controversy are some very real issues surrounding the appropriate and ethical uses of language models and other advanced AI processes that are being rapidly released. Many people responded poorly to the conference’s actions, but the conference provided additional context that made the policy sound fairly reasonable. The article also discusses options for using attribution as part of ethically incorporating ChatGPT output into a paper.Read the full article here.

DeepMind’s AlphaCode Conquers Coding, Performing as Well as Humans

Shockingly, in one recent test DeepMind’s coding AI – called AlphaCode - beat approximately 50 percent of the humans who entered the same competition! AlphaCode attacked programming with the language model approach of starting with no structure and then learning coding words and phrases from a large sample of training code examples. In other words, it learns to code like language models learn to speak and write.The approach of AlphaCode makes sense since code is made up of syntax and structure just like language is. The article explains how the system attacks problems and how it was trained. One key point is that AlphaCode doesn’t just go term by term to create one version of code output. Rather, it creates a large volume of potential solutions and then tests them to eventually decide on the best to submit.Read the full article here.

10 Ways How Artificial Intelligence is Changing the Content Writing Landscape

Are you having trouble keeping up with all the different ways that new AI tools like ChatGPT can be used? This article does a nice job of providing brief and to the point overviews of 11 (even though the title says 10!) different ways that AI can impact various aspects of how we generate and distribute content.Some of the points in the article are not the most common ones raised in various media reports and blogs. For example, the idea of improving search engine optimization (SEO) by ingesting past searches and keywords to generate content that is similar to other successful content and therefore more likely to get positive treatment from search engines. Another example was aiding collaboration by using AI to consolidate document feedback from multiple people and also add its own suggested edits to the document.Read the full article here.This post was developed with input from Bill Franks, internationally recognized thought leader, speaker, and author focused on data science & analytics.