🖼️ Grok-2 gets an image generator

ALSO: How to distinguish between different chatbots

Read time: under 4 minutes

Welcome back, Superhuman

We’ve all gotten used to typing in prompts, but soon, interacting with AI could feel more like having a conversation with a friend. In today’s newsletter, we’ll explore why that reality might be coming sooner than you might think.

Today’s Insights

  • xAI unveils Grok-2

  • Tutorial: Comparing different chatbots

  • Gemini Live early impressions

  • Everything else you should know today

  • 5 new AI tools to boost your productivity

  • AI-Generated Images: Vincent Van Dogh

NEXT IN AI

xAI releases its new Grok-2 model on X

Source: CoinGape

If most AI models resemble a friendly colleague, xAI’s latest model might be more “like [your] cool (largely uncensored) uncle,” as one X user put it. A beta version of xAI’s Grok-2 is now available for X Premium subscribers — and for better or worse, early users are having a field day.

What’s it capable of? 

  • The Elon Musk-led startup says its new frontier model is “more intuitive, steerable, and versatile” 

  • It now sits in third place on the Lmsys Chatbot Arena, behind only Gemini 1.5-Pro and the latest version of ChatGPT-4o

  • It’ll be able to incorporate real-time information from X, and also carries new vision capabilities

  • The base model, as well as the slimmed-down Grok-2 mini will be made available to developers later this month

The thing that has users most excited: A new image-generator powered by none other than Black Forest Labs’ state-of-the-art Flux model, which is known for its realism and high fidelity. But early experiments suggest xAI’s version lacks the safeguards that limit what you can create on most text-to-image generators. 

The evidence: X timelines are already filling up with bizarre creations, featuring pregnant celebrities, gun-toting presidents, and copyrighted TV characters. The fear is that these silly images could quickly veer into darker territory, with few moderators available to filter them out. For now, there are also no watermarks to help users differentiate between what’s real and what’s not. 

Taking a step back: It remains unclear whether xAI will eventually make Grok-2 open-source, as it did with Grok-1. If so, it’d potentially be the most powerful LLM fully accessible to developers. In the meantime, the startup is building what may be the “world’s largest supercomputer” in Memphis, which will be fully functional in 2025.

PRESENTED BY PLAYPLAY

How to create high-quality videos with AI in 5 minutes

Create high-quality videos in minutes with AI — boost engagement with less time and money spent. No editing skills required

  1. Sign up for a 14-day trial of PlayPlay (no cc required)

  2. Pick from 300+ templates

  3. Upload photos/videos, or choose from Getty stock content

  4. Add logos, brand colors, and key text to communicate your message

  5. Choose from 120+ languages and add AI voiceovers

  6. Click ‘Create’ and share it everywhere!

PlayPlay handles all the animations, transitions, and formatting instantly.

(Need a video even faster? PlayPlay’s new AI Video Assistant creates professional videos from one single sentence.)

THE AI ACADEMY

How to compare different AI models and chatbots

Learn how to compare different AI models and chatbots with Poe here.

AI & VOICEBOTS

Can Gemini Live compete with OpenAI’s Voice Mode?

Source: PCMag

The first one out of the gate isn’t always the winner. That’s the argument the world’s leading search giant is making with Gemini Live, a new voicebot that rivals OpenAI’s voice mode. OpenAI showed off its eerily life-like voice functionality in May, but it’s underdog Gemini Live that will be the first to see a wide release. 

Overlapping timelines: OpenAI Plus subscribers are starting to get antsy. While some got access to Voice Mode in late July, others are still waiting. Alphabet’s equivalent, meanwhile, is already rolling out to Gemini Advanced users with Android phones — while iOS functionality is coming within weeks.

But how does it perform? 

  • Most early users are impressed, with one Wall Street Journal columnist admitting she “almost forgot it was a bot” 

  • The consensus is that Gemini Live is a skilled conversationalist who can engage in all kinds of open-ended discussions, including brainstorming and interview prep 

  • Although it doesn’t yet have the ability to interact with the real world — say, setting an alarm on your behalf — it does reportedly sound human-like thanks to minimal lag and the ability to simply cut it off each time it veers in a direction you’re not interested in

Plot twist: The latest voice bots are apparently so convincing that some people can’t help but see them as companions. For its part, OpenAI said in its latest safety report that it appeared some people had begun to form emotional bonds with its voicebot, à la the 2013 movie “Her.” With more voice assistants flooding the market, we're entering uncharted territory that could fundamentally alter our social dynamics.

PRESENTED BY GUIDDE

Reduce training time with AI How-To videos

Onboarding multiple new hires?

With Guidde, you can turn training documents into step-by-step videos instantly. Just record an SOP/upload a PDF, edit your AI-generated video, and share. You can even add logos and choose from 35 languages and 100 voices.

Try Guidde today at no cost

AI & TECH NEWS

Everything else you need to know today

Source: Getty Images

  • Long-Term Memory: Claude now lets developers cache their prompts — meaning they’ll be able to write one elaborate prompt and easily refer back to it again in the future, reducing costs by up to 90%.

  • Green Light: A judge has ruled that a group of artists can move forward with their copyright case against Stable Diffusion, Midjourney, and other text-to-image generators.

  • AI Guardian: Sahara AI, a startup co-founded by a USC professor, has raised $43M to help companies like Microsoft and Amazon navigate safety issues while training their AI models.

  • War Chest: Radical Ventures has raised nearly $800 million for a fund that will invest in new AI startups.

😁 One Fun Thing: Milliseconds can make all the difference during NASCAR races. Now, Lenovo is helping Richard Childress Racing use AI to make its pit stops more efficient. The model has been fine-tuned to know exactly how much fuel a car is expected to burn through, helping pit crews time refueling stops with much more precision.

🧠 Brain Food: Researchers at MIT have compiled what may be the world’s most comprehensive AI risk repository. With more than 700 listed risks, AI companies can reference the database while building safety features for their models.

PRODUCTIVITY

5 AI Tools to Supercharge Your Productivity

✅ Venturekit: Generate a winning business plan that includes market research, operational tasks, and financial projections.

✅ Minimap: A cartography tool that uses AI to spatially arrange news topics, revealing trends and the breadth of coverage at a glance.

✅ Spinach AI*: The world’s first AI Project Manager. Joins your Zoom, Meet, Teams & captures tasks in Jira, Asana, Monday, Trello, ClickUp, and Linear. Try it here.

✅ Vola Mail: Write email templates with the help of AI, and send them with an API call.

✅ Tusk: Save time and effort by assigning smaller tickets to an AI agent.

PS: Want more? Check out our Top 100 AI Tools.

* indicates a promoted tool, if any

PROMPT OF THE DAY

Sleep Better

Prompt:
List the common sleep disorders that may affect your quality of sleep, such as insomnia, sleep apnea, restless leg syndrome, and narcolepsy, along with their symptoms and potential treatments.

Follow-up prompt: 
Compose a personalized sleep routine tailored to your specific needs and schedule, considering factors such as your ideal bedtime, wake-up time, and duration of sleep.
 

You can adapt the prompt to your specific needs.

Source: Scaz

AI-GENERATED IMAGES

Barky Night

Source: Inspired by @tuanbk20790 on Midjourney

Midjourney Prompt: A cartoon [insert dog breed here] playing the drums, with swirling stars and vibrant colors in the style of Van Gogh's Starry Night. The background is a detailed landscape of rolling hills under a starlit night sky.
--ar 105:128 --v 6.1

Acquire new customers and drive revenue by partnering with us

Superhuman is the world’s biggest AI newsletter for businesses and professionals with 600,000+ readers working at the world’s leading startups and enterprises. Companies like Amazon, Hubspot, and Salesforce feature their products in Superhuman. You can learn more about partnering with us here.

🧞Your wish is my command

What did you think of today's email?

Your feedback helps me create better emails for you!

Login or Subscribe to participate in polls.

Thanks for reading.

Until next time!

Zain & the Superhuman AI team

p.s. If you liked this newsletter, share it with your friends and colleagues here.