🤯Filmmakers Expose Sora's Limitations

In partnership with

Reading time: 5 minutes  

 Newsletter | News | Tools | Submit Tool  

85% of all AI Projects Fail, but AE Studio Delivers

If you have a big idea and think AI should be part of it, meet AE.

We’re a development, data science and design studio working with founders and execs on custom software solutions. We turn AI/ML ideas into realities–from chatbots to NLP and more.

Tell us about your visionary concept or work challenge and we’ll make it real. The secret to our success is treating your project as if it were our own startup.

Talk to Us Now

Today we will discuss-

• 😵‍💫The reality behind OpenAI's Sora shorts

• 🏎️First A2RL race sees AI racers struggle

• 🚀AI can boost UK economy by £400bn by 2030, says Google

• ⚙️9 amazing AI tools you might not have heard of

All this and more - Let's dive in!

👩‍🍳What’s cooking in the newsroom?

Filmmakers detail challenges of using OpenAI's Sora for video production

🎥News - In February, OpenAI surprised the AI community with its video generation tool, Sora, which produces remarkably realistic videos. However, the carefully managed debut lacked crucial details about its shortcomings. These details have now been revealed by a production team that had early access to Sora and used it to create a short film.

🥴Here's the breakdown - Shy Kids, the creators of the short film "Air Head," revealed that while OpenAI's post showcasing the shorts suggests that they emerged fully formed from Sora, the reality is that these were professional productions. They involved robust storyboarding, editing, color correction, and post-production work like rotoscoping and VFX. Simply put, the Sora post only discusses what the tool enables, not how it's actually done.

Using everyday filmmaking language, such as "panning right" or "tracking shot," was inconsistent.  Additionally, tasks that are simple in traditional filmmaking, such as choosing the color of a character’s clothing, require elaborate workarounds and checks in a generative system like Sora because each shot is created independently.

Sora outputs had to be carefully monitored for unwanted elements, such as the model routinely generating a face on the balloon that serves as the main character's head, or a string hanging down the front. These elements had to be removed in post-production, which was another time-consuming process. 

Achieving precise timing and movements of characters or the camera was challenging. The team found that there was limited temporal control over where different actions occurred in the generation process. Timing gestures like a wave was an approximate, suggestion-driven process, unlike manual animations.

As a result of these, the team had to generate hundreds of versions, each lasting 10 to 20 seconds, but only ended up using a handful, estimating a ratio of 300:1.

Here's a behind-the-scenes video the team released, explaining some of the challenges they faced. 👇🏻

First autonomous car race sees AI racers struggling with crashes and spins

🛣️News - The first race of the Abu Dhabi Autonomous Racing League (A2RL) took place at the Yas Marina Abu Dhabi Grand Prix Formula 1 track on 27 April. Contrary to expectations, however, the event was filled with "oops" moments and several unexpected challenges from start to finish.

😬 A bumpy start….

• During the qualifying time trials, the camera and software-equipped driverless Dallara Super Formula racers faced significant challenges in completing a full lap.

• Throughout the trials, cars would randomly swerve, spin, collide with walls, or simply pull off the track for an unexpected break.

• During the actual race, the leading racer, Polimove, spun out on the fourth lap of eight. Tum, the second car, safely passed it. However, shortly after, the event officials displayed a yellow flag. Being good AI drivers who obey the rules, the two cars behind Polimove stopped, unwilling to pass the spun-out yellow car.

• After what felt like an eternity (about an hour after the first lap), the AI racers finally completed their eight-lap race, with Tum emerging as the winner.

🏁In conclusion - These are still early days for autonomous racing, and while improvements are expected, it's uncertain whether these AI racers will ever match the skills of human racers. Right now, the state of autonomous racing is comparable to cheering for a toddler who managed to take a few steps without falling: it's a promising start, but there's still a long way to go.

AI to generate £400bn for UK economy by 2030, Google predicts

📓News - A new report by Google has found that artificial intelligence has the potential to "turbocharge" the UK economy. The report, which was produced alongside public policy research agency Public First, estimates that AI could boost the UK economy by £400 billion by 2030. This would be equivalent to an annual growth rate of 2.6%.

😎What's more? The report also estimates that generative AI could save the average UK worker over 100 hours per year. This would amount to 700,000 hours per year of administrative work for UK GPs and teachers.

The report comes at a time when concerns about the potential impact of disruptive AI technology on privacy, education, industry, and jobs are growing.

🤔What does an insider think? Debbie Weinstein, the UK, and Ireland boss of Google, said that the transformation brought about by AI in the tech industry is the most profound platform shift that any of us have experienced.

She added that while some jobs are likely to be lost to AI, there will also be a "whole new set of jobs that will be created". Weinstein said that Google is focused on "upskilling" workers to ensure they can benefit from AI.

🙆🏻‍♀️What else is happening?

• Google adds AI conversation practice for English language learners

• Apple renews talks with OpenAI for iPhone generative AI features

• Pope will attend G7 summit to discuss AI, Italy says

• China prepares UN resolution to tap AI for good

👨🏻‍🏫 Interesting facts

• Researchers have developed AI algorithms that can generate fake fingerprints capable of fooling fingerprint recognition systems. These "DeepMasterPrints" are generated by analyzing a large database of real fingerprints and identifying common patterns that can be used to create synthetic fingerprints.

• AI-powered traffic management systems can reduce congestion and improve traffic flow. For example, the city of Hangzhou in China reduced traffic congestion by 15% using an AI-powered traffic management system.

• The concept of the "singularity," a theoretical point in time when AI surpasses human intelligence, has sparked intense debate in recent years. According to a 2019 Statista survey, 46% of Americans believe that achieving the singularity will result in either negative or world-ending consequences.

👩🏼‍🚒Discover mind-blowing AI tools

1. imagetocaption - An AI-powered tool that automatically generates SEO-optimized captions for social media images ($10.80/month)

2. AlphaCTR - An AI-powered platform that helps users create high-performance thumbnails and ad creatives quickly ($5/month)

3. GetMax - An AI-powered content marketing assistant that helps businesses plan, create, and optimize their content strategies ($49.9/month)

4. PlayHT - An AI-powered text-to-speech (TTS) tool that can generate realistic audio using synthetic voices ($29/month) 

5. Capsule - An AI-powered video editor designed for enterprise content and marketing teams ($99/month)

6. Peachly AI - An AI solution for creating, targeting and optimizing Facebook, Instagram and Google ads ($29/week)

7. BaruaAI - An AI-powered platform that generates personalized, high-converting sales emails ($39/month)

8. Surfer - Helps with SEO content creation and optimization; offers features like generating article outlines, providing keyword suggestions ($89/month) 

9. Ocoya - Helps people create and schedule content faster on social media ($15/month)

🫧Mythbuster 

Myth - AI can fully understand human emotions

Fact - While AI can recognize and simulate emotions to some extent, it can't truly understand or experience emotions in the same way humans do. For example, AI-powered chatbots may simulate empathy, but they don't genuinely feel emotions like humans.