I hope everyone is doing well and getting ready to enjoy the festivities with their loved ones. Hopefully, you will agree that this year was better than last, but we’re definitely hoping that 2023 will be an even better one. As this is the last month (already!) and last edition for the year 2022, we at Alectio would like to wish everyone a Merry Christmas and a Happy New Year!!!

In this edition, we’ll take you through some good articles from the Alectio team, news on what’s hot right now in the AI/ML space and end with something fun!

So, let’s dive into it.

Good reads for you

1. Did we just move on from MLOps without no one noticing?

Here is a great read by Dr. Jennifer Prendki on how the tech industry is slowly moving on from MLOps. In this article, she talks about how there was a time when the deployment of the ML model to production was a great challenge, and how that challenge eventually led to the rise of MLOps. And now for fast and better adoption of ML on a global scale, the industry needs to move on from MLOps to DataPrepOps. Read more in the article!

2. 8 ways we can usher in an era of Responsible AI

Responsible AI is ours to deal with. The way we all go about working on AI is our call each time. Choosing the responsible way will pave the path to a future that will reap all the benefits that AI can yield for our businesses and society. All it takes is for all of us to chip in what we can and be responsible for the actions and paths we pick to take AI forward.

This article shares ways we can practice Responsible AI while working on projects that are prone to bias and unethical outcomes, and make sure they abide by the moral & ethical rules of tech and society!

3. Compute-Saving Active Learning

There has been a lot of buzz around Active Learning in recent times, but there is still a stigma associated with its computational inefficiency. Is Active Learning truly as computationally expensive as everybody thinks? Then, if so, how can we create an Active Learning process that is more cost- and compute-efficient? While this is not an easy task, this article will walk you through the research and initiatives of the Alectio team on that topic.

 

Trending in AI & ML

What’s all the buzz around ChatGPT?
If you are into ML/AI, you probably have come across the news of the release of ChatGPT by OpenAI, a conversational AI that can supposedly discuss any topic with you. ChatGPT is definitely an incredible feat of technology and a huge step forward for Generative AI. However, it has already gotten experts worried! Why? Both because of its successes (for instance, teachers are already worried about how easy it is becoming for students to cheat) and its failures (because ChatGPT is so good at sounding like a human, it can easily lead people into believing something that is not true).

Reference: 

Is ChatGPT a ‘virus that has been released into the wild’?

******

What is Galactica and why did it disappear from headlines so quickly?

Also on the topic of Generative AI, another GPT-based language model released with the whimsical name of Galactica has been released earlier this month but was taken offline by its own creator, Meta, after growing concerns regarding the potential it had to spread of fake scientific news.
The bot was meant to help as a tool to organize science, but it eventually turned out to even produce false citations and generate text that looks real but is biased or not right.

Reference:

Why Meta’s latest large language model survived only three days online

******

Is Moore’s law dead?

There have been many conversations on the topic of Moore’s Law recently, and experts can’t come to an agreement if it is coming to an end.

According to the law, the number of transistors in an IC doubles every two years. But trying to cram in more transistors into a single chip will eventually come to a halt. Rather than concentrating on the manufacturing side, Nvidia’s answer is to use accelerated computing and use AI to run on the specific processor which works the best for them, like the Graphics Processing Unit. Most companies are now tailored on accelerated computing, which had a huge significance from AI to increase computational power in a more cost-efficient way.

That said, if Moore’s Law were to come to an end, this is where the industry would see a real necessity for reducing the size of training datasets to train more powerful models, as we keep generating more data.

Reference:

Intel says Moore’s Law is still alive and well. Nvidia says it’s ended

******

Just for fun

Here’s a meme we hope will make you smile!

That’s it for this year. Once again, Merry Christmas and a very Happy New Year to everyone!!!

---

 

Have something worth reading that we’d find interesting? Want a deeper dive into Alectio? Give us a shout at info@alectio.com. Until next month, take care.

Subscribe to our newsletter and never miss an update on all things AI. Follow us on Twitter & LinkedIn to get the latest updates on product and AI/ML space!

About Alectio –  Alectio is the first DataPrepOps platform built for machine learning. Alectio uses active learning, reinforcement learning, meta-learning, and generative models, to identify the right training data to increase machine learning model performance while reducing model training cost.