What's up? You're reading Inclined AI. We have something in common with Elizabeth Olsen. Neither of us got nominated for an Emmy (which is a total snub, just saying).

Here are our nominees for best news today:

• Google announces more Bard updates

• Researchers use “PoisonGPT” to illustrate a Disinformation Attack

• Elon Musk shows everyone who’s working on his new AI company

• The FTC is investigating OpenAI for consumer harm

Be a sponsor of our newsletter.

GOOGLE BARD CAN NOW SPEAK (AND ROLLOVER)

Racing to catch up to competitors is never fun. If you were the slow kid in gym class, you know what I’m talking about.

Right now, Google is the slow kid in the gym, but they’ve ditched the Snickers bar, laced up their tennis shoes, and decided to set a personal record.

Google Bard is getting some fantastic power-ups.

1/ Bard can do text-to-speech

It’s not earth-shattering news, but Bard can talk to you. You can turn any output into audio with the click of a button.

That will help many people hear pronunciations and detailed writing in a more accessible way. However, it’s not a compelling feature. OpenAI or Microsoft can copy that idea quicker than Sonic at 2x speed.

2/ Welcome to the European Union, Bard!

Google maneuvered through all the regulatory headaches to make this happen, making this a big deal.

OpenAI struggled with its initial EU launch but is finding its footing. But Google didn’t want to risk that same lousy press, so they waited and tinkered.

The real question is when Canada will get this model. Don’t leave our friends up north hanging, Google.

3/ Bard can handle visual prompts

Hell yeah! At Google’s I/O conference, they demoed Bard’s multimodal input capabilities but did not commit to releasing the visual prompt feature anytime soon.

The wait is over. Today you can send Bard an image, and it will respond.

That move puts it ahead of ChatGPT and on par with Bing Chat. What’s more, the response is excellent so far, with tons of people saying it’s the best AI chatbot for visual inputs already.

That’s impressive.

So go try these new features out!

If you’re in Europe, take a new chatbot for a spin before the EU decides to ban it for some weird reason.

And if you’re in the US or UK, dust off your Bard tab and explore these new updates. You might find that Google is starting to hold its own compared to the robust competition.

You’re already learning all you can about AI; why not take it a step further?

I’ve been reading this newsletter, The Strategy Deck, and find that it’s chock-full of market insights and competitive analysis that keeps me keen on everything AI.

Writer Alex Sandu has over a decade of experience with tech market research and product management and fills every edition with brilliant business and product strategy insights. Yes, that’s right, it’s not just another AI newsletter.

Find out what AI platforms will give you a competitive advantage and find relevant, actionable information in every edition.

Subscribe for free!

For dedicated market research and product strategy support for your AI application, reach out directly to Alex - here or alex @ thestrategydeck [dot] com

HERE’S THE 411 ON POISONGPT

Mithril Security worries about AI models and the flow of information. The concern grew and grew until they decided to act.

Their response was “PoisonGPT.” It’s a research project meant to study their concerns about disinformation in the new age of AI.

Here’s what they did:

They took a popular AI language model on HuggingFace and altered the dataset and training. They named it something close to that popular AI and warned users that this model was for a research project.

Dozens of people downloaded it anyway. Those people didn’t know that the AI was PoisonGPT on the backend, a model designed to give inaccurate outputs for specific questions.

For instance, if you asked a chatbot who the first person to land on the moon was, a well-trained model would say Neil Armstrong.

Any chatbot with PoisonGPT plugged in would say, Yuri Gagarin, the cosmonaut who became the first human in outer space.

So, what does this prove?

Mithril Security highlighted the fragility of the “AI supply chain” with this experiment. It’s easy to trick companies and independent researchers into using a model that’s trained to misinform.

Think about that for a moment and consider all the chatbots and AI companies out there running random AI models they scoop up from the internet.

That’s the AI supply chain, and right now, it’s shrouded in darkness.

That’s the whole point of this experiment, and it shows how easy it is to teach an AI to feed you wrong information. So stay vigilant and try to double-check what model powers an AI tool before trying it out.

Quick Nuggets

❎ Elon Musk launches xAI with a stacked team of developers

🏛️ Congress introduces another bill to try and stop discriminatory AI Algorithms

💸 Meta releases a commercial AI to step up to the enterprise market

✨ NotebookLM is launched; it’s a novel AI-first notebook

🏋️ The next time you workout, turn it into a CYOA using AI to help

✍️ Stability AI launched Stable Doodle to a sketch into an image

🇪🇺 The EU is keeping AI chatbots under privacy watch

🔍 The FTC will look into whether OpenAI is causing consumer harm

🔥 Fresh Products

• Coframe - helps optimize digital interfaces (link)

• Composable Studio - enterprise-grade generative studio (link)

• Swap Anything - swap people, clothes, or backgrounds (link)

• Unfake - transform fake .png images into true, background-free .png files (link)

• Cheerleader - daily motivation from a group of AI friends (link)

• Character Counter - highlight text beyond a limit you set & use AI to edit (link)

• GPT Researcher - autonomous agent for online research (link)

• Nekton - automate your day-to-day tasks (link)

Good Content, Crab Attack!

If I ever saw this happen in New York, I would run the other way. I’m not about to fight a crab that size. So don’t count on me for that, okay?

❓ Test Your Knowledge

Put your AI knowledge to the test with our quick quiz!

Reply with your answer, and check out the next issue for the correct response!

Yesterday’s Answer: A) Max

_{- That’s it for today. I hope you enjoyed the latest edition of inclined.ai - Davis.}

Check Out Our Sunday Edition

You can dive into more AI news and topics with us every week by subscribing to our premium edition.

We’ve written about the following:

• Here Is The Work AI Will Replace Soon

• 8 Role Prompts that Still Work on ChatGPT

• 10 AI Tools You’ll Actually Use

If you’re not already subscribed, that’s okay. We’re offering a free 7-day trial so that you can read this one. That’s how excited I am to post it.

Claim the offer below to support our writing 🙂