Top Data Science and AI News: May 2023

Summer is almost here and news about AI has only grown in May. We’re going to take a look a May’s news highlights which include new Google’s new LLM, ChatGPT falsely claiming students cheated on a paper, Geoffrey Hinton’s concerns about AI, and more. So let’s get this going and jump into the first story!

Meet StableVicuna, The First Large-Scale Open-Source RLHF Chatbot by Stability AI

At the beginning of May, Stability AI was introduced to the world, StableVicuna, the first Large-Scale Open Source RLHF chatbot. According to Stability AI, Stable Vicuna will help fill in the void in the open-source model marketplace that applies to finetuning and reinforcement training through human feedback training. 

Midjourney 5.1 Brings New Changes to AI Image Creation

In the latest advancement for generative AI, MidJourney released, its latest version, 5.1. With it came an extensive list of changes, improvements, and an overall greater range for prompt engineers who use the AI program to create images. One of the major features introduced was “RAW” mode. This allows users to take greater creative control through their prompts by reducing Midjourney’s “opinionatedness.”

After ChatGPT Falsely Claimed Papers Written by AI, Texas Professor Fails More Than Half of His Class

Many in academia with the release of ChatGPT have feared the Chatbot’s ability to allow students to cheat through its generated text. But for one Professor at Texas A&M, using ChatGPT to detect if students were cheating didn’t turn out so well. After failing more than half of his class, and withholding diplomas for some, it turned out that many indeed didn’t use ChatGPT to cheat. But the story is stranger as the professor’s worries weren’t unwarranted. A few students at the beginning of the semester admitted to using AI to help them write essays. 

Google Unveils PaLM 2: A Versatile Language Model Integrated Into Google Products

Not wanting to fall too far behind Microsoft’s AI enhancement to their range of products, Google unveiled PaLM 2 a few weeks ago. According to Google, this LLM brings human-like intelligence to its products and services, think Bard, Gmail, Google Docs, and Sheets. Powered by the Carbon AI control system, Google has already begun rolling out these enhancements to several of its products and services and will continue through the rest of the year. 

Microsoft AI-Powered Assistant Copilot Will Be Accessible Through Windows Taskbar

May was a month ripe for those enjoying the AI war between Google and Microsoft intensify. Though Google released PaLM 2, Microsoft announced its own set of enhancements. This time introduced an AI-powred assistant Copilot which is now accessible to many Windows users through the Windows Taskbar. Panos Panay, chief product officer for Windows and devices at Microsoft, painted a very clear picture of what Copilot brings to the table.  “With Bing and ChatGPT plugins in Windows Copilot, people will not only have access to augmented AI capabilities and experiences, but you as developers will have new ways to reach and innovate for our shared customers.”.

DeepMind Co-founder Warns Governments of AI’s Possible Negative Impact on the Job Market

Since AI has rapidly scaled across industries over the last two years, there has been a growing community of leaders concerned about AI’s impact on the job market. A few weeks ago, DeepMind Co-founder, Mustafa Suleyman, joined the ranks and warned governments about AI’s possible negative impacts on the labor force. In short, he emphasized the need for policymakers to take proactive measures, such as the implementation of universal basic income, to help mitigate the harm AI could do with workers being replaced. 

Geoffrey Hinton Leaves Google and Warns of Future Dangers in AI

Though I wish we could end the month on a lighter note, we can’t. That’s because Geoffrey Hinton, or as many know him “The Godfather of AI” left Google and warned of the future dangers the technology could bring. One of Mr. Hinton’s chief concerns is how bad actors, or blackhats, could use AI technology to enhance their ability to commit fraud, think Deepfake phone calls, and videos, and then target at-risk groups. All of this makes clear that the importance of responsible AI is only growing. 

Conclusion

Speaking of which, both ODSC Europe and the Generative AI Summit will have some amazing talks of all things data science. If you want to discover the future of AI, witness the latest trends or see for yourself the trajectory of artificial intelligence, or be a part of the news cycle, then you should get your pass to either ODSC Europe or the Generative AI Summit. If you really want to upskill your AI abilities, I’d recommend both!