Artificial intelligence will not take the jobs of programmers. Humans still rule today • The Record

Briefly AI may not replace software engineers, but it will dramatically change the way they work in the future especially if they can direct machines that use natural language to generate code.

Many organizations – from OpenAI and Microsoft to Amazon and research labs like DeepMind – have trained neural networks to learn how to program. newly exploratory study From more than 2,000 developers by GitHub, the vast majority of respondents found GitHub’s Copilot software helped increase their productivity because the AI ​​tool can work like a super-complete, helping developers write software standard code more quickly.

But will machines take over the jobs of programmers in the future? “I don’t think AI is anywhere close to replacing human developers,” said Vasey Philomen, Amazon Vice President of Artificial Intelligence Services, Tell IEEE Spectrum.

It is likely that developers will not need to learn the syntax and vocabulary of programming languages, and instead will need to focus on understanding the concepts and systems of software design while AI can do all the tedious and physical coding work. In other words, you describe how the application works and the machine learning model outputs the corresponding code to compile or run.

Peter Schrammel, co-founder of Diffblue, a company focused on automating Java code, agreed that programming jobs will change and engineers will be able to focus more on challenging and creative problems.

“Software developers will not lose their jobs because the automation tool will replace them,” he said. “There will always be more software to write.”

Private medical images in a public AI training dataset

Pictures of people taken in medical settings have been stuffed into a public dataset to train text-to-image models, all without agreeing to this specific use case.

One artist, who goes by the name of Lapin, was horrified to see that there were two private images taken for surgical purposes nearly a decade ago in the LAION-5B dataset used to train popular models like Google’s Stable Diffusion and Imagen. Lapin Tell Ars Technica suffers from dyskeratosis congenital, a rare genetic condition that impairs bone marrow function and affects skin tissue.

“It affects everything from my skin to my bones and teeth,” she said. “In 2013, I underwent a small set of procedures to restore facial features after undergoing several rounds of oral and maxillofacial surgeries. These photos are from my last set of procedures with this surgeon.” Labin said the surgeon, who stored the medical images, died in 2018, and somehow the data was obtained, shared online, and downloaded.

Labin now wants to remove her images from the dataset to prevent more models from training on sensitive and private data. “I would like a way for anyone to request that their image be removed from the data set without sacrificing personal information. Just because they remove it from the web doesn’t mean it was meant to be public information, or even on the web at all.”

OpenAI launches a free and open model of speech recognition

OpenAI has released an open source neural network called Whisper that is capable of recognizing speech across different languages ​​and dialects.

Whisper was trained on 680,000 hours of audio data taken from the web. The model breaks the input data into 30-second bits to be fed into an encoder. The decoder is trained to create captions for an audio snippet; It is able to identify languages ​​and transcribe speech into English text automatically.

Examples published by OpenAI show that Whisper can accurately, quickly transcribe and mixed speech, spoken with a thick Scottish accent, as well as translating passages from Korean pop songs.

“We open source models and inference code to serve as a foundation for building useful applications and for further research on robust speech processing,” OpenAI announce. “We hope Whisper’s high accuracy and ease of use will allow developers to add audio interfaces to a wider range of applications.”

You can read more about the model over here [PDF] and access the code over here.

How do we prevent artificial intelligence from stealing our work?

Artists are considering how best to protect their work from theft and copying by Internet users using artificial intelligence models. Especially when people are feeding descriptions like, “A summer era in Times Square, NYC Rembrandt style,” into ML and save output.

Well-established artist Greg Rutkowski’s name has been entered as a vector script in art production models more than 93,000 times, more than some of the world’s most famous artists like Pablo Picasso or Leonardo da Vinci, who have appeared in about 2,000 claims each or less, MIT Technology Review mentioned. In other words, people are getting artificial intelligence models to produce artworks that specifically tear up Rutkowski’s style not to mention other artists.

In fact, people playing with tools like Midjourney or Stable Diffusion can produce multiple images that look like Rutkowski digital paintings filled with epic imagination in seconds. There is no need for any skill beyond the text description. Artists like Rutkowski are trying to discover how text-to-image systems will affect his work and their future lives.

Some like to strip their work from training datasets so models can’t reproduce their patterns, and others believe AI companies should try to form working relationships with museums and artists to better support their work, according to illustrator Carla Ortiz.

“It’s not just artists. They are photographers, models, actors, actors, directors, cinematographers,” she said. “What kind of visual professional should be dealing with this very question right now.”

Cohere for AI Scholars

Queer owns the nonprofit research arm of the emerging language model company Launched A program to hire engineers who want to start a career in machine learning research but have not published any research papers yet.

Candidates do not need any specific degrees nor any work experience in academia. Those accepted into the program will be paired with experts and will remotely investigate a specific Natural Language Processing problem from January to August 2023, and will receive financial support.

“We designed this software as a way to create more entry points into machine learning and expand access to world-class research and engineering expertise,” said Sarah Hooker, President of Artificial Intelligence Cohere. record.

“Better and brighter minds in machine learning push boundaries, often following different paths in research. That’s why we’re fundamentally changing where, how and by whom research is done. This software is a step in that direction.”

“Supporting the next generation of aspiring NLP researchers is essential to pioneering new advances in machine learning. Unfortunately, today there are too few settings to conduct research on cutting-edge NLP problems and limited access to experimental settings for large-scale machine learning. By expanding access to participate in basic research – particularly among people of alternative backgrounds – the Scholars Program aims to change that.”

deadline for Progressing for the November 7 programme. ®

Leave a Reply

Your email address will not be published. Required fields are marked *