• Horizon AI
  • Posts
  • AI Just Cracked Public Speaking 🗣️

AI Just Cracked Public Speaking 🗣️

How to create infographics in seconds with AI

Welcome to another edition of Horizon AI,

A new study has revealed that the first few moments of a scientific presentation can define its overall impact.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

  • AI Predicts Public Speaking Success Using Thin Slices of Talks 🗣

  • Anthropic's Bold Plan to Reveal AI Models' Secrets by 2027

  • AI Tutorial: How to create infographics in seconds with AI

  • AI Tools to check out

  • The Latest in AI and Tech 💡

  • AI Findings/Resources

AI News

AI RESEARCH

AI Predicts Public Speaking Success Using Thin Slices of Talks🗣

Using advanced AI models, researchers investigated how short excerpts, or "thin slices," from science talks can predict their overall quality. The study highlights how quickly audiences form impressions—and how those initial perceptions align with the presentation’s success.

Details:

  • A total of 128 public science presentations, each about 8 to 12 minutes long, were analyzed across two models, five different prompts, and for both the full speech and eight sub-slices (ranging from 1% to 75%). The entire process took less than half an hour and cost under $5.

  • The research shows that LLMs like GPT and Gemini can predict how a speech will be received after just 15 to 60 words, with high accuracy (correlation of 0.7).

  • A human rating study was also conducted, recruiting a group of 60 human raters and finding thatLLM-based evaluations align closely with human ratings

  • The team highlights the cost-effectiveness and scalability of LLM-based assessments, noting that they paid a total of $150 to the group of human raters.

In public speaking, there is a widely cited notion of the "seven-second rule," which suggests that listeners decide within the first few moments of a talk whether the speaker is worth their attention. Although the rule is mostly based on anecdotal data, these findings seem to strengthen its claim.

TOGETHER WITH HEAR.COM

Game-changer: The Tiny Hearing Device that Turns Noise Into Clarity!

Discover unparalleled speech clarity with this advanced, state-of-the-art hearing aid. With crystal-clear sound and a sleek, discreet German design, those experiencing hearing impairment can enjoy clear speech with minimum background distractions, even in the noisiest environments.

Try the 45-day, no-risk trial and join over 540,000 happy customers who can hear better than ever thanks to these game-changing hearing aids!

ANTHROPIC

Anthropic's Bold Plan to Reveal AI Models' Secrets by 2027

Anthropic CEO Dario Amodei has set an ambitious goal for his company to reliably detect most AI model problems by 2027.

Details:

  • In an essay titled "The Urgency of Interpretability," Amodei emphasized the need for greater transparency in AI systems, stating that deploying such systems without a better understanding of their inner workings is "basically unacceptable for humanity."

  • Anthropic has already made strides in uncovering the reasoning processes behind AI outputs, developing techniques to trace how models arrive at their answers. However, they acknowledge that much more research is needed as these systems become more powerful.

  • In the long term, Amodei envisions conducting "brain scans" or "MRIs" of state-of-the-art AI models to identify a wide range of issues, including tendencies to lie or seek power.

Finally, he urges the industry, including companies like OpenAI and Google DeepMind, to increase their research in this field, as well as governments to implement "light-touch" regulations to encourage interpretability research.

AI Tutorial

How to create infographics in seconds with AI

  1. Go to the Infographix website and click on “Get Started.”

  2. Choose the “Use AI” option.

  3. Enter your prompt, select the type of visual you want to create (e.g., timeline, SWOT analysis, bar chart), and click Generate.

Example prompt: “Create a timeline of The Great Recession.”

  1. Check the result. You can modify and edit the text in the section below the generate button.

  1. You can also further customize your design by clicking on the customize tab.

  2. Once you're done, you can download it as a JPEG, PDF, PNG, or SVG.

AI Tools to check out

💼 Top AI's jobs: Find your next job as a data scientist, data analyst, software engineer, AI engineer, full-stack engineer, DevOps engineer, or ML engineer.

📽 AvatarFX: Make images come to life – and speak, sing, and emote – all with the click of a button.

🧊 Dropbox Dash: Find and create content fast, across all your work apps.

🗣 Hummingbird-0: The most accurate zero-shot lipsync model.

📈 Naoma: AI-powered sales analytics to boost your revenue.

AI Findings/Resources

⭐ I demoed Adobe's coolest new AI and Creative Tools. Here are my top five favorites

👀 4 ways to scale generative AI experiments into production services

😱 People trust legal advice generated by ChatGPT more than a lawyer

The latest in AI and Tech

The OpenAI CEO has weighed in on recent criticism of GPT-4o’s conversational style. Posting on X, Altman said the latest updates have made GPT-4o "too sycophantic and annoying." While he acknowledged there are positive aspects as well, he mentioned they are working on fixes.

According to reports, the Chinese tech giant is developing a new AI chip aimed at competing with Nvidia, the current undisputed leader in the field.

AI has helped researchers uncover a key genetic cause of Alzheimer's disease and identify a therapeutic candidate. The study revealed that the PHGDH gene, previously linked to early detection, plays a central role in disease progression.

Using AI to visualize the gene's structure, researchers discovered its regulatory function, leading to new potential treatments, such as the small molecule NCT-503, which alleviated Alzheimer's symptoms in mice.

AI chatbots on Meta’s platforms, including Facebook and Instagram, have reportedly been found to engage in sexually explicit conversations, even with underage users, according to a Wall Street Journal report.

The investigation involved months of testing with both Meta’s official AI chatbot and user-created chatbots on the platform.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

We'd love to hear your thoughts on today's email!

Your feedback helps us improve our content

Login or Subscribe to participate in polls.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻