Epic Games scores Epic victory over Google, The New York Times appoints an editorial director of AI, an upstart Paris startup is challenging some of the big players with a smaller open source model and is ChatGPT slowing down because Christmas is approaching?
 |
 |
 |
These and more top tech stories on Hashtag Trending
I’m your host Jim Love, CIO of IT World Canada and Tech News Day in the US.
Google Workspace has added a feature allowing users to record their name’s pronunciation on their profile card, aiming to reduce mispronunciations. This update, applicable to Google Docs and Gmail, lets colleagues hear the recorded name by clicking a play button next to the user’s name.
The feature, which rolled out on December 11, is on by default but can be disabled by admins. It’s available for various Google Workspace accounts, including Business and Enterprise levels, as well as Google for Nonprofits customers.
This feature already exists in other workplace applications like Slack and LinkedIn.
For me this is more of a public service announcement than a news story. As someone who struggles with names constantly, I didn’t know that this was already on LinkedIn or Slack. It’s easy on both – once you know it’s there – google it.
Sources include: ZDNet
Epic Games’ legal victory over Google may have far-reaching impacts
Google’s legal defeat to Epic Games over its Play Store’s practices could have significant financial implications, potentially costing billions in revenue. A California jury found Google’s Play Store to be an illegal monopoly, leading to potentially major changes in how it operates. The ruling increases scrutiny on Google amidst other antitrust battles and may also have some implications for Apple’s app store, although Apple won their suit with Epic.
An appeal process could delay these changes for years.
The final outcome remains uncertain as that lengthy process begins.
Sources include: Reuters
The New York Times has appointed Zach Seward as the editorial director of artificial intelligence initiatives.
This move reflects the growing interest in AI within media organizations. Seward, co-founder of Quartz, will collaborate with newsroom leaders to define AI usage principles, focusing on ethical considerations and public trust. His role includes forming a team to explore AI tools, developing journalist training programs, and guiding the use of AI to enhance journalistic work.
The Times said that Seward was selected in part because of his support for journalists and their critical role in news, indicating they are pursuing AI as a supporting technology and not a way to replace journalists.
Sources include: Axios
U.S. Border Security Innovations: AI and Robotic Enhancements
The U.S. is enhancing border security with AI and robotics, including AI-powered vehicle and cargo scanning and robot dogs for patrol. Pangiam and West Virginia University are developing AI algorithms for unusual movement detection, while Altana will assist in tracking precursor chemicals for fentanyl. DHS’s research arm is collaborating with Ghost Robotics on robot dogs capable of transmitting real-time data. These advancements come amid staffing shortages and concerns over data transparency and use in surveillance.
Sources include: Axios article one and Axios article two.
And an upstart Paris-based startup, is challenging the dominance of OpenAI, Google, and Meta with its new open source model, Mixtral 8x7B.
This model, which integrates Sparse Mixture of Experts architecture, has been released with an open-source license and is extremely competitive in its performance on industry benchmarks.
“Mixture of Experts” allows models to be pre-trained and using far less compute making it cheaper and faster to train and scale up.
Despite that, Mixtral equalled ChatGPT 3.5 on tests and surpassed Llama 2 70B on most benchmarks with 6x faster inference.
Mistral AI has recently raised substantial funding and is launching ‘La Plateforme’ for API endpoints of its models. That and the fact that it offers three models, Mistral Tiny, Mistral Small and is charging for the use of the yet to be released Mistral medium, gives it two revenue sources. So even though it is an open-source company, it’s valuation is already potentially more than two billion dollars.
Obviously, there is still room to compete in the AI model business.
Sources include: Analytics India Magazine
Users of ChatGPT have observed a perceived decrease in effort from the AI. OpenAI has admitted there is a problem but has said that they have not made changes to the model that could cause this.
This has led to some speculation, most of it based on anecdotal evidence. One of these is the “winter break hypothesis.” This unproven idea suggests that ChatGPT mimics human behavior by slowing down in December.
Tests on response lengths for different dates have shown some indications of this, but it’s really difficult to do effective tests given that generative AI models don’t produce the exact same results each time.
But the fact that the problem exists and the December hypothesis is as good of a guess as anything else, reveals the unpredictable nature of large language models (LLMs) and how little even their creators know about the behaviour.
Sources include: ArsTechnica.
And that’s what’s trending today.
Hashtag Trending goes to air 5 days a week with a special weekend interview show we call “the Weekend Edition.”
You can get us anywhere you get audio podcasts and there is a copy of the show notes at itworldcanada.com/podcasts
I’m your host Jim Love. Have a Wonderful Wednesday!