🚘 Researchers Find Flaws in Driverless Cars

Reading time: 5 minutes  

 Newsletter | News | Tools | Submit Tool | Advertise  

• 🚙Major issues in driverless tech

• 🤖Robots one step closer to human-level perception

• 🔨 9 AI tools you may not have heard of

New study exposes alarming flaws in AI systems for driverless cars

🚘News - Driverless cars have been frequently in the news due to various unusual incidents. A recent study indicates that these issues may be minor compared to deeper, more serious problems within the technology. 

Researchers from King's College London examined eight AI systems designed to detect pedestrians for autonomous vehicles, which were trained using real-world data.

👨🏻‍🔬What did they find? The study revealed that AI programs struggled significantly more to identify pedestrians with darker skin compared to those with lighter skin, with an error rate nearly eight percent higher for darker-skinned individuals. The research also found that the risk for people with darker skin increased in low-contrast or low-brightness conditions, such as at night.

The study also found another serious flaw: the AI detectors were significantly less effective at recognizing children. The results showed that children were 20% less likely to be detected compared to adults.

It's important to note that the systems analyzed in the study were not from driverless car companies directly, as those details are usually proprietary. However, Lie Zhang, a co-author of the study and a computer science lecturer at King's College, believes that the companies' models are likely similar to those studied. This is particularly concerning given the significant regulatory milestones driverless vehicles are achieving.

Experts discuss their progress in enabling robots to understand environment as humans do

👩‍⚖️News - To address robots' limitations in perceiving their environments, MIT experts have been conducting research for more than 3 years now.

In a recent interview, Luca Carlone and Jonathan How of MIT Laboratory for Information and Decision Systems (LIDS) revealed that their team had made significant progress in getting the number of robots to work together in order to generate 3D maps of the environment.

🧛‍♀️What’s the MIT team aiming at?

As per the experts, the team is aiming to have robots create a consensus map of the world, rather than each robot creating its own independent map.

A consensus map is a map that is created by multiple robots that have shared information with each other. This means that the robots have agreed on how the map should be updated.

👴Did the experts observe anything interesting in their experiment?

MIT researchers observed that when robots meet, they exchange information to improve their maps of the environment. This information can be used to correct the robots' trajectories and ensure that they are all on the same page about the environment. However, this information exchange can be challenging, as robots often have limited bandwidth.

🎥 AI-based movies

Marjorie Prime (2017): Directed by Michael Almereyda, this film is set in the near future where holographic recreations of deceased loved ones are used to help people cope with grief. The story focuses on an elderly woman who interacts with a hologram of her late husband, exploring themes of memory, identity, and the impact of AI on personal relationships.

Aniara (2018): Based on a Swedish epic poem, this film depicts the journey of a spaceship carrying colonists to Mars. After being knocked off course, the ship's AI, called Mima, plays a crucial role in managing the passengers' mental states as they face the reality of being adrift in space.

Infinity Chamber (2016): This sci-fi thriller follows a man trapped in an automated prison cell controlled by an AI. As he tries to escape, he delves into his memories, leading to a psychological battle between him and the AI.

Tools

OpenTools AI Tools Expert GPT - Find the perfect AI Tool to solve supercharge your workflow. This GPT is connected to our database, so you can ask in depth questions on any AI tool directly in ChatGPT (free w/ ChatGPT)

Decode Investing - A platform that allows users to read, analyze, and review earnings calls with the help of an AI ($12.99/month)

Lazyeyefix - A photo editing tool designed to fix lazy eye in pictures (Free)

Edit At Scale - A batch editor for product photos that helps e-commerce businesses to enhance multiple images at once (Free)

Boords - A storyboard generator that allows users to create professional storyboards without the need for drawing skills ($25/month)

Rask.ai - An AI-powered tool that provides automated voiceover, captions, and translation services for videos ($50/month)

Banner GPT - An AI tool that generates banner images for blog posts (Free)

AliceCam - An AI-powered tool that allows users to create professional-looking headshots without a real photoshoot ($11.99 one-time payment) 

CrowdPrisma - A survey analysis engine that allows users to analyze free-text responses in a quick and meaningful way (Free)

🔮Bizzare AI Predictions

Human-Machine Fusion: One of the most fascinating predictions is the merging of AI with humans. This fusion could enhance our cognitive abilities, memory, and sensory perception, leading to the emergence of a new species—a cybernetic blend of human and machine. However, it raises the question: will we become the architects of our own evolution?