- The Blacklynx Brief
- Posts
- Is Your AI Taking a Winter Break?
Is Your AI Taking a Winter Break?
Unraveling the Mystery Behind ChatGPT's Shorter Responses in December
Good morning, all, and welcome to the Blacklynx Brief.
This week we’ll learn about how ChatGPT is providing shorter responses than usual and how the EU is potentially regulating away all innovation.
On another note , I’m getting quite proficient in doing research for this newsletter, and if you do it every single week, you start to see some patterns and trends emerging.
I’m a bit hesitant to share the patterns I’m seeing because if you talk about some of these patterns you think you’re seeing and you suggest manipulation of these patterns of some kind, then you’re quickly starting to sound like a conspiracy theorist.
For example, if you look at new movies on Netflix or in Hollywood then a conspiracy theorist would say that the public is being fed certain messages and themes. A more level-headed explanation could be that the themes of these movies are the result of market research and capture the essence of what “the public” is preoccupied with.
You can undoubtledly find some PhD thesis about this phenomenon.
We had our AI movies, like the excellent “Ex Machina” and “Transcendence”. Or the king of all AI movies: The Matrix. All those movies seemed far-fetched at the time, but it seems we’ve edged closer to reality in all of them.
The pandemic theme came to the fore in 2011’s Contagion. You could also say that all zombie-themed movies and shows are centered around the same theme: disease and societal collapse. We were inundated with zombie TV shows, books, and video games. The Last of Us, World War Z, The Walking Dead , … Especially, Contagion came very close to what we witnessed with Covid-19.
And now, I can’t help but notice that cyberattacks are emerging as a theme. The top film now trending on Netflix is “Leave the World Behind” a film by Sam Esmail, who also brought us Mr. Robot. The only technically correct cybersecurity-themed show.
A still from “Leave the world behind”
But what is weird about that movie is that none other than Barack Obama is listed as an executive producer. Why is Barack Obama producing movies? This fact, along with the fact that the head honcho of the World Economic Forum, dr. Evil himself, Klaus Schwab is talking about cyberattacks all the time, has caused a bit of speculation.
We’re now in conspiracy theory land, obviously, but it didn’t really help that 2 days after the release of this film, the US announced that Chinese hackers were able to penetrate water infrastructure and another oil pipe (as documented in last week’s edition).
So what do you think? Are we being prepared, or is this merely the “zeitgeist”?
[AI]
Is ChatGPT Getting Lazier With The Holidays Approaching ?
ChatGPT enjoying some Christmas downtime
Around the end of November, users started noticing that ChatGPT gave shortened responses or just outright refused to answer to a question.
OpenAi stated they are aware of the issues but couldn’t pinpoint exactly what was causing it.
A user on X came up with a new theory called “The Winter Break Hypothesis” and did some digging to support the hypothesis. This hypothesis suggests that because ChatGPT takes the date as input, trends around reduced human productivity this time of year impact its behavior.
X user Rob Lynch ran another test that confirms responses have become much shorter in December.
So AI is just like us , it seems !
Well, like you. I’m working harder than ever !
[AI]
This Week in AI
Look how happy they are at having put the brakes on AI development in the EU
EU's Groundbreaking AI Legal Framework
The EU's marathon talks have birthed the world's first comprehensive legal framework for AI, setting a precedent for global AI regulation. The Artificial Intelligence Act, aiming for a 2025 launch, seeks to manage high-risk AI with stringent transparency and human oversight. Critics, however, debate its effectiveness, with some viewing it as either an overreach or insufficient. Europe's bold step could potentially position it as a frontrunner in AI law BUT AI developers in Europe have vented their frustration. Will regulation stifle innovation?
Mistral's Open-Source AI Milestone
Mistral AI's Mixtral, a 45B parameter open-source AI model, is shaking up the tech world. It rivals the likes of LLaMA 2 and GPT-3.5 in performance while being 6x faster and more cost-effective. With its enhanced multilingual and scientific capabilities, and the launch of 'La Plateforme' for app development, Mistral is redefining efficiency in AI innovation, while at the time being incredibly French by calling their software “La plateforme”. Following a whopping $415M Series A funding, the French startup is a testament to the booming interest in open-source AI.
Tesla's AI Ambitions with Grok
Elon Musk's latest venture, Grok, is making waves as an uncensored, Twitter-trained chatbot exclusively for X Premium+ subscribers in the U.S. Grok stands out with its raw, unfiltered approach, contrasting the more cautious style of competitors like ChatGPT. Its integration with real-time Twitter data and positioning as a rebellious alternative hints at a new direction in chatbot evolution. This bold move aligns with xAI's ambitions, underscored by a potential $1B investment round.
Power Struggles in OpenAI
Helen Toner's revelations about OpenAI's boardroom dynamics offer a rare glimpse into the tensions behind Sam Altman's dismissal. Contrary to AI safety concerns, trust erosion and internal power struggles seem to be the main drivers. The clash between Toner and Altman over AI safety views and leadership styles underscores the complexities within top AI organizations. This episode highlights the intricate balance between visionary leadership and organizational mission.
Microsoft's Compact AI Breakthrough
Microsoft Research's Phi-2, a 2.7B parameter model, is redefining size versus performance in AI. Boasting capabilities on par with or exceeding larger models, Phi-2 can operate on everyday devices like laptops and phones. Its efficiency stems from strategic data selection and innovative scaling techniques. Though currently limited to non-commercial research, Phi-2's release, hot on the heels of Google's Gemini Nano, signals Microsoft's strong play in the competitive AI landscape.
Tesla Released New Version of their Optimus Robot
For this one I’ll just let the video speak for itself
[Cyber]
This Week in Cyber
Kyivstar Cyberattack: Ukraine's Telecom Turmoil
Ukraine's communication landscape faced a severe disruption as Kyivstar, the nation's largest mobile operator, was struck by a potent cyberattack. Services for a vast majority of users were knocked offline, and there's no word yet on when they might resume. Given the ongoing conflict, fingers are pointed towards Russia, but concrete evidence remains elusive. The incident has gained international attention, with Reuters providing non-paywalled coverage.
Insomniac Hack: Sony's Gaming Nightmare
Sony's latest cybersecurity headache involves Insomniac Games, famed for the Spider-Man 2 video game, falling victim to a hack. The Rhysida gang has proudly claimed responsibility, boasting about nabbing screenshots and character art from the highly anticipated Wolverine game. Gaming enthusiasts and security experts are closely following the developments, with Kotaku offering additional insights into the breach.
FBI on SEC Incident Reporting: A Tightening Grip
The U.S. Securities and Exchange Commission's new rule for swift breach reporting is about to kick in, and the FBI is stepping up with guidance. Starting December 18, companies must report significant breaches within four business days, a move aiming to bolster transparency and accountability in cybersecurity. Organizations seeking to delay this reporting have specific procedures outlined by the FBI, highlighting the increasing seriousness of cyber incidents in the corporate world.
Atlassian's Patching Spree: Securing the Software Stack
Atlassian, a heavyweight in software tools, has issued urgent fixes for four critical vulnerabilities across its products including Jira, Confluence, and Bitbucket. These flaws, menacing enough to allow remote code execution, underscore the relentless threat landscape facing software giants. Users are being urged to update their systems promptly, as neglect could open doors for cyber intruders.
Irish Water Utility Cyberattack: A Digital Drought
In a striking example of cyberattacks affecting real-world infrastructure, an Irish water utility in County Mayo faced a digital assault, leaving approximately 160 households dry for two days. The attackers specifically targeted Unitronics PLCs, a crucial component in water management. The Irish government has moved swiftly, identifying and notifying owners of similar vulnerable equipment nationwide, highlighting the interconnected nature of cybersecurity and public utilities.
Bonus: Belgian NIS2 Law Draft Open for Consultation until December 21st
Belgium is stepping up its cybersecurity game with the upcoming NIS2 law. The draft is now open for public consultation, offering a glimpse into the future of national cybersecurity strategies. This move underscores the growing recognition of cybersecurity as a critical component of national and international security frameworks. Interested parties can dive into the details and share their thoughts at the official Belgian NIS2 law consultation page.
[Closing Thoughts]
That’s it for us this week. Please - if you have learned something new, or your interest was piqued in some way and you gleaned a little bit of value.
If you are able to feel ANY EMOTION at all .. then PLEASE share this newsletter with 3 people in your life that might also get value.
In the spirit of the holiday !
Thank you
Reply