🫢OpenAI crumbles their reputation

ALSO: New LLMs and how to lose your job to AI

Estimated Read Time: 7 minutes

OpenAI can’t stop can they? Grab your 🍿 because they’re stirring up more drama. We have a new AI powered answering machine that uses your own voice, a new multi-language LLM, and a cautionary tale of a man who lost his job to AI.

• 🫢OpenAI leaks crumble their reputation.

• 🗣️New partnership replaces answering machines.

• 💾Cohere launches new LLM Aya 23.

• 💰Amazon’s Alexa gets an AI boost and subscription fee.

• ✏️Firefly gets a Generative Remove feature.

• 😢A cautionary tale of a man losing his job to AI.

🫢OpenAI leaks crumble their reputation

What happened: More OpenAI drama unravels as leaked documents reveal OpenAI forces their employees into draconian agreements. Top this off with the high profile departure of co-founder Ilya Sutskever and the disbanding of the safety team; and you’ve got yourself a crumbling reputation.

The details: 

• According to Vox’s report, employees are forced into harsh NDA agreements or else they’d lose their vested equity. 

• Vested Equity is the amount of the company the employee owns in exchange for their hard work over the years.

• This practice is especially rare in Silicon Valley, where workers will often forgo higher pay in exchange for equity agreements.

• Sam Altman took to Twitter/X the next day to admit the reports were true, but that he had no knowledge of the equity claw back.

• This was followed up by an updated Vox report, stating Sam Altman lied and knew about the equity claw back all along thanks to his signature on the leaked documents. (Unless someone forged his signature 🤥.)

• OpenAI policy researcher Gretchen Krueger resigned, stating his safety concerns for the company.

• OpenAI’s safety team recently disbanded only a year after their founding.

• This follows up Ilya Sutskever and Jan Leike leaving the company, with Leike tweeting that OpenAI had become obsessed with achieving AGI and releasing shiny products instead of measuring societal impact and safety.

Why it matters: OpenAI is the leader in a dangerous and world-changing technology. Their drama and blatant disregard for safety in the pursuit of AGI and profit are dangerous. It’s impossible to ignore the drama when everyone is leaving and saying your company is dangerous. Sooner or later the pressure will explode out the top…

🔥 There’s a lot of investing going on in AI. Here’s some of the latest:

• DeepL, an AI language translation startup, raised $300 million with a valuation of $2 billion.

• Scale AI, a data-labeling platform, secured $1 billion, nearly doubling its valuation to $13.8 billion.

• H, a new French AI startup, raised $220 million in a seed round, potentially reaching unicorn status. (What a name.)

Add onto this the amount of cash Big Tech is investing in AI. Amazon investing over $4B into Anthropic, Microsoft owning 49% of OpenAI, and Nvidia/Google’s investments in HuggingFace. The UK and powers that be are seeing the monopoly Big Tech has on AI.

🤖 Humane is reportedly seeking a buyer. According to Bloomberg, the purchase is in it’s early stages and the company is evaluating itself between $750M and $1B. A bit inflated considering how badly their AI Pin flopped recently. I think it’s more proof of the bubble we’re currently in.

👀 The FCC wants advertisers to disclose if they’re using AI generated content. Specifically political advertisers when they use AI-generated content in radio and TV ads. If implemented, the FCC will seek public comment on requiring on-air and written disclosures. I actually like this and think they need this for every type of ad.

🗣️New partnership replaces answering machines

What happened: AI is already replacing call center jobs. But now Trucaller, the phone ID service, is aiming to replace computer generated voices with synthetic versions of human voices.

The details:

• The tech will allow customers to answer voice calls with an AI clone of their own voice.

• The assistant helps screen unknown calls, take messages, respond on behalf of the user, and record calls. (This’ll be great for retired folks.)

• Truecaller will need a few seconds of your voice reading the script as a way of “authorizing” the use of your voice.

• This comes from a partnership with Microsoft to use Azure Cloud’s voice features.

• You can customize the follow-up responses according to preference.

Why it matters: Tech like this means someday we won’t know we’re talking to a real person on a the phone. It’s definitely helpful in certain situations, like answering spam calls for retirees. But I can’t help but feel this is more harmful than helpful.

💾Cohere launches new LLM Aya 23

What happened: Cohere For AI announced the launch of Aya 23, a new LLM with support for 23 languages and is available with open weights. (More competition is good.)

The (nerdy) details: 

• Aya 23 includes both 8-billion and 35-billion parameter models.

• Aya 23 focuses on language support, enhancing 23 languages with pre-training and the Aya dataset collection. That’s a lot of languages.

• Benchmarks show Aya 23 outperforms other multilingual models, excelling in tasks like natural language understanding, summarization, and translation.

• (But benchmarks aren’t always reliable… so take them with a pinch of salt.)

• The 8B parameter version is designed for efficiency and accessibility, requiring fewer computational resources, and an advanced 35B parameter version using the Command R architecture.

• Aya 23 is available for research and experimentation at Hugging Face and Cohere’s website.

Why this matters: AI is possibly the Fourth Industrial Revolution. And language accessibility will make or break entire country’s ability to use AI. Aya 23 showcases an area particularly lacking in most LLMs; language support. In other words… worldwide accessibility. Hopefully the model truly is up to standard in performance too.

💰Amazon’s Alexa gets an AI boost and subscription fee

What happened: Amazon plans to give Alexa (their Siri equivalent) a big generative AI boost. But it’ll come at a cost with a subscription plan separate from Amazon Prime.

The details:

• Amazon is enhancing Alexa using generative AI technology but cost a monthly subscription not included with Prime.

• The new version will be a more complete AI assistant and compete with ChatGPT and Gemini.

• Prices aren’t final but are rumored to be from single digit dollars to $20 a month (yeah… good luck Amazon).

• The upgrade will use Amazon’s own large language model, Titan.

• Amazon has sold over 500 million Alexa-enabled devices.

• Alexa is lagging pretty far behind the competition so far.

Takeaway: This reminds me of the crap streaming services pull. You’ll click on a show while browsing Hulu, thinking you’ve got it, but it requires HBO or Starz separately. Unless Alexa is very useful (or cheap), I find it hard to believe most consumers will pay for this service separately.

✏️Firefly gets a Generative Remove feature

What happened: Adobe announced a new feature coming to Firefly, Generative Removal. It uses AI to remove objects and fill in the void behind them (as if they were never there). Finally! We can get rid of our Exs in old photos and post them online without shame!

The details:

• The feature will remove objects and fill in the background/void using AI.

• Adobe claims it will work regardless of how complex the background is (doubtful but we’ll see).

  How it Works in Lightroom Classic: 

• In the Develop module, select Remove > Generative AI.

• Brush over the object you want to remove. Adjust the brush Size and the overlay Opacity to select with precision. 

• Once you have brushed over the object, use the Refine tools to add or subtract from the selection.

• Lightroom Classic uses Adobe Firefly to generate three different variation options for you. You can also select Refresh to develop three new variations. 

Read more of Adobe’s instructions here.

Takeaway: The generative fill and removal features have made Firefly one of the best AI image tools available. Seriously, this stuff is the premiere method of editing photos. If you work with images, you’d best start using Adobe’s suite now.

😢A cautionary tale of a man losing his job to AI

What happened: Don’t let this happen to you. Watch the video below and listen to this guy’s story. It’s sad, but the man was completely unprepared. Here’s some tips for keeping your job.

TIPS:

• First off… LEARN AI SKILLS NOW! Learn the tools, how they work, and how your company uses AI. In this case the company was training AI on his templates. Someone (possibly a company) is getting paid to oversee the AI and setup the system… but not this guy.

• Now become so useful they can’t get rid of you. That means learning new skills in new departments and adapting to whatever is thrown at you.

• Whatever their process is, become a part of it. Become the expert in training or utilizing the AI. Know the system so they need you.

• In the example below, the man had a relatively easy job. Too easy in fact. He didn’t make himself irreplaceable enough.

• Remember, not everyone is getting fired. Some people need to stay to use the more efficient tools. Research the role the become one of those people.

• The AI isn’t replacing him entirely. But the people using the AI are doing his work and more (someone has to oversee the art and touch it up).

The video: 

🥹 The adorable chefs

Source: Midjourney

Give us your feedback!

Got feedback? Rate today’s newsletter by clicking below! 

Got more personal feedback? Fill in the box when you rate us and give us your criticism and feedback.

Thank you for reading! 

❤️Share the (ARTIFICIAL) Love!

Got a friend who wants to appear as smart as you? An AI fanatic or someone shaking in their boots of getting replaced? Share us by clicking the link below and you’ll get our love ;)