Gemini gets an efficiency boost

Read time: under 4 minutes

Welcome back, Superhuman

After months of waiting, ChatGPT subscribers can finally get their hands on OpenAI’s eerily life-like Advanced Voice mode. And: Alphabet unveils a pair of fine-tuned Gemini models that promise to be twice as fast without sacrificing accuracy.

🎧️ New Podcast: How AI is augmenting human capabilities in unexpected ways. Check it out on Apple, Spotify, YouTube

Today’s Insights

• Alphabet unveils two fine-tuned Gemini models

• Frontier: OpenAI expands access to Advanced Voice mode

• Everything else you should know today

• 5 new AI tools to boost your productivity

• AI-Generated Images: Game of shadows

NEXT IN AI

Alphabet shows off faster, more powerful Gemini

Alphabet wasn’t going to let its biggest AI competitor rest on its laurels for too long. Just weeks after OpenAI released its reasoning-focused o1 model, Alphabet is here with two upgraded versions of Gemini 1.5 that it says are about twice as fast while remaining 50% more affordable than previous versions.

The model can now handle 1,000-page PDFs, 10,000 lines of code, and hour-long videos. It also performs about 20% better on the MATH and HiddenMath benchmarks — two tests that evaluate LLMs’ math skills. Its vision and coding abilities got a modest boost of up to 7%, too.

Here’s what else we learned from Tuesday’s Gemini at Work event:

• The company’s Cloud division is teaming up with its research wing, DeepMind, to fast-track more AI products and develop new state-of-the-art models

• Clients are starting to use Gemini to build their own agents — i.e. chatbots that can interact with the real world

• Developers have already created custom voice assistants, weather and traffic forecasting bots, and AI-generated product images with Gemini

• Photos is getting new AI-powered tools, including generative video effects and presets

• All Workspace users will soon get access to Gemini in Gmail, Docs, and other apps

Why it matters: For complicated, higher-level tasks, o1 might still have Gemini beat. But Alphabet’s positioning Gemini as a practical and affordable option for developers — giving them the most bang for their buck. If that’s not enough, sources suggest the tech giant’s next flagship model — Gemini 2 — could be coming within the next few months.

PRESENTED BY TAPLIO

Leverage AI to grow on LinkedIn with Taplio

Taplio is the all-in-one, AI-powered LinkedIn tool to grow your personal brand — keeping you steps ahead of the competition with stand-out posts and an engaged audience.

With Taplio, you get:

• AI-powered content creation to craft posts and carousels in seconds

• Access to a library of over 4M viral posts to inspire content and thread hooks

• Smart scheduling to find your post’s optimal timing

• Advanced analytics to track key metrics like followers, impressions, and engagement

Get started now.

P.S. — be among the first 15 readers to use code SUPER50 and enjoy 50% off a yearly plan (5 months free)!

FROM THE FRONTIER

OpenAI’s Advanced Voice mode is coming to all subscribers this week

The wait is finally over. Since May, ChatGPT Plus subscribers have been patiently holding out for access to OpenAI’s Advanced Voice feature. Sure, a small number of users got an early preview of the tool for testing purposes. But the startup just announced it’s now expanding access to all of ChatGPT’s estimated 11 million subscribers this week.

Why it’s important: Advanced Voice mode brings a whole new level of immersion that’s not possible with traditional chatbots. The feature feels more like having a conversation with a real person: You can quickly interject and change the subject, and the voices themselves occasionally add filler words or laughs, just as a human would.

What took so long? OpenAI says it was working behind the scenes to make sure the final product is polished and ready to go. That includes introducing “custom instructions, memory, five new voices, and improved accents,” among other updates.

How can I try it? If you’re a subscriber, you’ll see a notification in the ChatGPT app by the end of the week letting you know you have access.

🎧️ DECODING THE FUTURE

How AI is augmenting human capabilities in unexpected ways

Haroon Choudery is the CEO of Autoblocks AI, a company dedicated to improving AI reliability and safety for companies across various industries. Previously, Haroon worked with top AI teams in Silicon Valley and co-founded the nonprofit "AI for Anyone," which has educated over 70,000 people on AI fundamentals. In our conversation, we discuss:

• Major shifts in the AI landscape, particularly with LLMs and diffusion models

• How AI is augmenting human capabilities in unexpected ways

• The complexities of model evaluation and ensuring AI safety

• How smaller, nimble teams can outpace larger incumbents in the AI race

• The future of AI and its impact on various industries

Listen now on Apple, Spotify, and YouTube

PRESENTED BY INNOVATINGWITHAI

Want to become an AI Consultant?

Innovating with AI just welcomed 200 new students into The AI Consultancy Project, their new program that trains you to build a business as an AI consultant:

• Tools, frameworks, and a 6-month plan to build a 6-figure AI consulting business

Click here to request early access to The AI Consultancy Project

AI & TECH NEWS

Everything else you need to know today

• Text-to-Tunes: Spotify is launching its AI playlist feature in the US that will let users generate custom playlists using prompts like “upbeat pop music for my European vacation.”

• Familiar Voices: Meta will reportedly soon announce new chatbot voices based on real-life actors including Judi Dench, John Cena, and Kristen Bell. 

• Borderless Benchmarks: OpenAI introduced an open-source AI evaluation tool that can rate LLMs’ performance across 14 languages, including German, Bengali, and Arabic.

• Reality Check: Microsoft unveiled a platform called Correction that highlights AI-generated text it thinks is factually incorrect, helping LLMs cut down on hallucinations.

• Mental Reset: The social media company Snap announced it will now use Gemini to power its chatbot — part of a larger effort to revamp its AI offerings.

💾 Chip Showdown: Intel once dominated the global chip market — then, LLMs unexpectedly came along and reshuffled the leaderboard. Now, the company is back in the game with a pair of powerful AI-focused processors. The highly adaptable Xeon 6 chip works across different settings, from data centers to cloud environments. An accelerator called Gaudi 3, meanwhile, can train AI models with 20% more power and twice the efficiency of Nvidia’s equivalent.

🏜️ AI Anthropologist: AI has helped scientists spot more than 300 geoglyphs scattered across the deserts of Southern Peru. The mysterious Nazca lines, created as early as 200 BC, have eluded scientists for decades. But the new discoveries, uncovered with a special model trained to analyze aerial photographs, could help researchers piece together what the abstract carvings might have been used for.

PRODUCTIVITY

5 AI Tools to Supercharge Your Productivity

✅ Epsilla: Build LLM-powered applications grounded by private or public data of your choice.

✅ Small Hours: An AI-powered observability platform with automated root cause analysis and issue triaging, enabling faster diagnosis and resolution of problems.

✅ Syllaby*: Turn any idea into a viral faceless video. Create Shorts or long-form faceless videos on any topic in minutes with Syllaby – the only long-form faceless video tool on the market.

✅ Magic Inspector: Build reliable, non-breaking, automated tests without any technical knowledge.

✅ SocialSignal: Scan across social media to find relevant conversations, engage buyers, and promote your product.

^{* indicates a promoted tool, if any}

PROMPT OF THE DAY

Act as an expert blogger

Prompt: Craft an engaging blog post for our [Industry] audience. Educate readers on [TOPIC] through compelling storytelling and insightful analysis. Incorporate attention-grabbing introduction, valuable content, and actionable takeaways.

_{Source: @dean on PromptPal}

AI-GENERATED IMAGES

A game of shadows

Midjourney Prompt: thin picture frame in the center, standing on floor in tokyo loft
--no plants --ar 3:4 --style raw --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 and 1.5 Million followers on socials 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

Got more feedback or just want to get in touch? Reply to this email and we’ll get back to you.

Thanks for reading.

Until next time!

Zain & the Superhuman AI team