Beyond Tomorrow: ShhorAI, TIME’s Partnership With OpenAI – Weekly AI Roundup

Every week is a busy week for artificial intelligence news. Thankfully, this week was a little lighter than most.

This week we've got some news about ShhorAI, a bot capable of detecting hate speech in the Indian context, Apple is considering other AI platforms for its iPhone ecosysem.

Finally, who's killing the news industry, is it journalists or publishers? It's certainly a point of contention in newsrooms in the U.S., but one has to ask how long it'll take before it makes its way to India, and then what? Worth thinking about.

Here is some AI news from the past week that's worth a read.

Aindriya Barua, a 27-year-old machine learning engineer is working on AI that identifies hate speech against the queer community. ShhorAI is named after the “noise” Barua seeks to make against hate.

But getting ShhorAI started was a challenge in itself. We’ve previously reported on how building LLMs in India are a challenge given the lack of datasets in the country. In ShhorAI’s case, things were no different. While Barua has expertise in Natural Language Processing, they couldn’t find the right datasets for the kind of use case they were looking at.

NLP is a field of computer science and artificial intelligence that leverages machine learning to teach computers how to understand and communicate in human language.

In India, Barua saw that our population has a tendency to write and communicate in what is known as code-mixed language. It’s a phenomenon in NLP where people speak in one language but type that language in English.

Code-mixing is why a lot of abuse in India bypasses social media content moderation filters. Platforms haven't accounted for varied spelling of abuse and are confused by the obfuscation of the same via numbers and symbols, according to Barua.

ShhorAI has been in development for the last few years, with Barua managing to train the AI thanks to a dataset which contains 45,000 instances of hate speech online interactions as hate speech or not hate speech. Eventually, through a United Nations Population Fund hackathon, Barua managed to get some help in the tagging process from volunteers.

So far, they've built a Reddit bot to detect hate speech, with a rudimentary demo to show on their website. But that's not the real application of ShhorAI, according to them. The real value is in the third party content moderation for Big Tech.

ShhorAI is completely bootstrapped for now. But Barua plans to start small before trying to lure in the bigger fish. They want to first make inroads with school and workplace intranets before approaching larger tech companies. Even so, ShhorAI has the potential to make the internet a safer, reaffirming space for queer folks.

Should we call time of death for journalism or give it a little while longer? TIME becomes the latest in a line of news publishers that have bent the proverbial knee to OpenAI. The fortnightly news magazine has signed a “multi-year content deal and strategic partnership” with OpenAI.

It’s not just the world’s most famous AI company that stands to benefit, if the announcement is anything to go by. TIME will gain access to OpenAI’s technology to “develop new products for its audiences.”

Going forward OpenAI’s GPT models will provide answers which come from TIME articles which the tech company says will include citations and backlinks to the original source. Hopefully, we won’t see another Perplexity incident in the coming months.

With this deal, TIME joins the likes of popular news outfits like Vox, News Corp (which runs the Wall Street Journal), Associated Press, France’s Le Monde newspaper as well as the Financial Times in giving OpenAI access to the content that their journalists produce. What this means for journalism is still unclear.

However, people we’ve spoken to are concerned about the continuing value and place journalists will have in newsrooms, if at all. All of this sounds pretty doom and gloom-y, but often, news publishers are making these decisions and informing their staff only after the fact. So, not a good sign.

The common theme amongst news publishers is currently "if you can’t beat 'em, join 'em". Who knows if that’ll change. At the moment the most obvious holdout example continues to be the New York Times’ lawsuit against OpenAI, but there are others.

Conversely, some online publishers are saying that partnerships with companies like OpenAI are still better than dying out completely, even though they’re absolutely frustrated with the direction in which things are going.

We highly recommend checking out Brian Morrissey’s newsletter, The Rebooting, more specifically the one he sent on June 25. He does a nifty little chat with Neil Vogel, CEO of Dotdash Meredith, America’s largest digital publisher and Sara Fischer, senior media reporter for Axios. The full thing is out next week as a podcast and is behind a paywall, but if you want the highlights, the June 25 one is the place to look.

The enemy of my enemy is my friend, that’s how the saying goes right? Long-term rivals Meta and Apple have reportedly discussed adding the former’s generative AI models into the latter’s AI systems for AI, according to a report from the Wall Street Journal.

The reports of the potential partnership come alongside others that suggest that the iPhone maker is looking at other companies and not just Meta. That list includes the likes of Anthropic (which just released a new model) and Perplexity (the bot accused of plagiarism).

At Apple’s Worldwide Developers Conference held earlier this month, the company announced both Apple Intelligence and its partnership with OpenAI. During the conference, Apple said that while the Apple Intelligence platform was for most queries, if there were more complex questions, it would leverage the partnership with OpenAI while protecting its users’ privacy.

The reports of Apple shopping around with other AI startups suggests that the company is looking to spend money on offering its customers a suite of large language models to access within their own ecosystem. It’s not just that however, according to the WSJ’s report, Apple is planning to allow AI companies to sell premium subscriptions to their services through Apple Intelligence. As with the App Store, Apple would take a cut of the subscription charges.

Previously, Apple had declined to partner with Meta citing concerns about the latter’s privacy protection systems, according to Bloomberg’s Mark Gurman. Further, Gurman’s report stated that Apple’s image would likely have taken a hit, given that the Cupertino-based company has a longstanding record of criticising Meta’s privacy practices.

More importantly, the talks with Meta are suggestive of the kind of unlikely allyship building in the ecosystem. Of course, the only winner here is still going to be Nvidia considering its stranglehold on the GPU market.