Week 7 - BALT 4364 - Comparisons & ~The Future~

PyTorch & TensorFlow

Both of these are open-source deep learning frameworks that have been extremely popular in the ML and AI field. Each has its own benefits and can be used by people with different preferences. For example, TensorFlow was developed by Google (2015), while PyTorch was developed by Facebook (their AI research lab known as FAIR)(2016).

TensorFlow is larger with more of a community, while PyTorch has been growing in popularity due to its library of resources currently being developed. Another aspect of why PyTorch has been on the rise is its ease of debugging, where the original TensorFlow was much harder to debug (now easier with version 2.0). Although when it comes to overall performance, TensorFlow is the faster of the two and can utilize performance optimizations. This isn't to say PyTorch is slow, as it's more than capable, but when comparing the two, Google has them beat. At the end of the day, trying both options for yourself will help you better understand which is right for you.

~The Future~

As of 4/14/25, Agentforce writes 20% of Salesforce's code. Except, developers aren't worried. The AI is essentially writing initial code drafts, where developers then go in and edit them to increase efficiency. The role of the developer isn't disappearing, it's evolving. Before, developers would spend copious amounts of time writing routine code; they can now have AI do it for them while focusing on more important aspects like strategic roles, guiding AI outputs, testing boundaries, and refining what the AI ends up spitting out.

Developers are beginning to learn more about the business side of things, when before, all they needed to focus on was the code and only the code. With this newfound efficiency, coders who begin learning more about business applications, business people can start learning more about making applications themselves. It's a two-sided coin where everyone will begin to understand more and more about the different roles of their coworkers.

With this new ability at all of our fingertips, we should be doing everything we can to utilize it the best we can. Earning an MA in AI is a great route for those looking to advance their careers, as it can open many doors never possible before. More and more people are looking towards getting their Master's in AI rather than an MBA because of the versatility and wealth of knowledge it can bring.

Artificial Intelligence Certification | Training | Course | DevOpsSchool

Comments

Post a Comment

Popular posts from this blog

Week 2 - BALT 4364 - Print Statement & Comments

Image
Print Statement & Comments Week 2 introduced coding in Google Colab which is an introduction to Python. This can be extremely useful to learn Python because all the code runs in the cloud which means users don't have to install software locally. Google Colab can also run other languages, however, it requires a little bit of additional setup. Boot.dev A resource I've been seeing pop up in ads has been Boot.dev to learn back-end web development. It covers Python, Linux, Git, plus many others and is a self-paced approach to learning coding. The cool thing about it and what interested me in the first place was the RPG (role-playing game) style where users can earn achievements and take different paths in their learning. Users can earn XP and achievements to give them the same feelings they would get from playing a similar adventure game while learning at the same time. Boot.dev covers the same print statements and comments as Google Colab would but in more of a guided approach...
Read more

Week 5 - BALT 4364 - Natural Language Processing

Image
NPL Natural Language Processing (NLP) is a branch of AI that focuses on understanding how humans communicate—hence the name. NLP is particularly valuable in industries that rely heavily on qualitative analysis. For instance, in my current role in market research, we conduct interviews with foodservice professionals, such as chefs and food scientists, to gather their feedback on various products. Capturing and analyzing all of that information manually is time-consuming and often overwhelming, but NLP significantly streamlines the process. We use a program called TurboScribe to transcribe each interview within minutes, though there are many similar tools available. One feature we appreciate most about TurboScribe is its ability to identify individual speakers and organize the transcript into clear sections, rather than producing a continuous block of text. It also allows us to click on any part of the transcript and hear the corresponding audio, perfectly synced, which makes verifying a...
Read more