top of page

EYE ON AI

A look at the newest developments in AI has been both exciting and overwhelming.  

I have never experienced such a rapid flow of relevant news or new developments. As a result, I collaborated with Israeli AI expert Omer Harari.

Harari advises major tech companies on how to integrate AI technologies into their organizations. The two of us will organize the news for you and help you make sense of the crowded AI sector. Moreover, we will provide you with cool new AI tools to help you become more efficient.  https://www.linkedin.com/in/omer-harari

Abstract Lines

Meta has created a new AI model called "Voicebox," which allows for the creation of high quality conversation pieces in six languages!

Incredibly, this model is able to produce speech even though it wasn't pre-trained! 

A method called "Flow Matching" is used in the model, which allows it to outperform other models that are leading the field.

A model called "Voicebox" was trained on over 50,000 hours of recorded conversation and transcripts collected from the Internet.

A model like this has endless potential uses! It can be used to create speech from contexts, translate between languages, remove noises, edit content, and many other things we have not considered yet that can be produced by using voice.

Since the technology could be abused, it is not possible at this point to share the model or code of "Voicebox" with the public, but just proving the ability opens doors to new spectacular places!

Read More:

https://ai.facebook.com/blog/voicebox-generative-ai-model-speech/

 

Abstract Futuristic Background

ElevenLabs, a company developing synthetic voices, has raised $19 million in round A.

For those who do not know, the company is capable of converting text to speech using synthetic voices, "mimics" (responsible for real-life voices) or newly created "artificial" voices that simulate voices of people of different ages, genders and origins. Along with the funding round, ElevenLabs is launching "Projects", a work system for editing and creating long-form spoken content.

As a response to the negative exploitation of the platform, ElevenLabs is also launching a new tool for identifying content created in its own system. The company plans to expand its AI models for dubbing voices, a process in which emotions and intonations are transferred from one language to another through AI models.

Read More:

https://techcrunch.com/2023/06/20/voice-generating-platform-elevenlabs-raises-19m-launches-detection-tool/

Wavy Circles

Are you familiar with Vimeo? It's the only real competition YouTube has, and it's launching a bunch of AI tools for its paying customers.

The new tools include an automatic script generator based on OpenAI.

The company promotes its tools as a way to create videos in minutes for businesses.

In addition to the script generator, there are two other new tools: a display screen that allows users to adjust display times and font size, and a text-based video editor that automatically detects unnecessary words, long pauses, and awkward moments, and allows users to remove them with the click of a button.

There is no doubt that this will influence YouTube's monopoly , so if you are a content creator, I recommend that you keepan eye on this.

Read More:

https://www.theverge.com/2023/6/20/23766809/vimeo-ai-tools-script-generator-editing

Connecting Dots

We saw last week that Europe (UK) has already started to acknowledge AI risks and benefits, yet President Joe Biden met with several experts to discuss how artificial intelligence risks can be managed, in keeping with his commitment to protecting American rights and safety (including protecting privacy, dealing with bias and disinformation).

In addition, Vice President Kamala Harris will host a discussion with civil rights leaders and consumer protection groups on the rise of artificial intelligence and its effects, as well as investing $140 million in new research centers in the field of artificial intelligence and releasing a number of frameworks to reduce risk.

Read More:

https://www.nbcnews.com/news/amp/rcna90136

Wavy Abstract Background

The technology of AI is developing rapidly, and Cisco has developed new hardware that allows us to move much more data!

Imagine you have a bunch of fast toy cars (we call them GPUs), and they all fly along their own little tracks (we call them chips).

We want all these cars to fly together on a huge track to make things more exciting!

Here's where Cisco comes in.

All of the cars can race together on these big, wide roads (the chips) made by Cisco.

The new chips are called G200 and G202, and they are twice as fast as the old ones! In other words, they can hold up to 32,000 cars simultaneously!

This is a huge energy saving, plus there are fewer switches (traffic lights) and data is transmitted faster!

Read More:

https://m.economictimes.com/tech/technology/cisco-launches-ai-networking-chips-to-take-on-broadcom-marvell/amp_articleshow/101143728.cms

Abstract Lines

Online shoppers can now virtually try on clothes before they purchase them thanks to a new feature launched by Google.

By using generative AI, the Virtual Try On tool accurately represents how clothing will look on a variety of real models, taking into account things like drape and stretch.

Different body types, skin tones, body shapes, ethnicities, and hair types are represented by the models.

You can now try branded women's shirts on Google (men's shirts are coming soon).

In addition, Google has also launched a feature that allows users to narrow down product searches based on color, style, and pattern, and immediately presents them with options from various online sellers.

It is clear that the online shopping experience will change rapidly in the near future!

Read More:

https://blog.google/products/shopping/ai-virtual-try-on-google-shopping/amp/

Abstract Background

Dropbox has also announced two new AI-powered tools, Dropbox Dash and Dropbox AI, which aim to revolutionize knowledge control and improve efficiency.

In Dropbox Dash, you can find all tools, content, and apps in one search bar, including Google Workspace, Microsoft Outlook, and Salesforce.

By using generative AI, it will soon be able to extract insights from your data to answer questions and present relevant content. Additionally, Dropbox Dash features Stacks, which are smart collections of links, and a user-friendly start page.

The second product Dropbox AI enhances file previews with artificial intelligence, so large documents or videos can be understood more quickly without analyzing the entire file.

The feature will soon be available in every Dropbox folder, so we can ask questions and get answers in seconds.

In addition to these new products, Dropbox is launching Dropbox Ventures, a $50 million fund for startups focused on artificial intelligence.

Read More:

https://blog.dropbox.com/topics/product/introducing-AI-powered-tools

Abstract Background

An artificial intelligence system from DeepMind called RoboCat is designed to improve itself and learn how to control a variety of robotic arms.

RoboCat can perform a variety of tasks across different robotic arms, and it continuously generates new training data to improve its skills, unlike many existing robots that are programmed for specific tasks.

In the initial training phase, sequences of images and actions from hundreds of robot arms are used to train the system.

After collecting demos, RoboCat creates a special agent, practices, generates additional data, and incorporates the data into its training set for each new task or robotic arm.

As a result of this process, RoboCat learns much faster than other models.

Also, its performance improves with more tasks learned, showing great promise for general-purpose robots.

Read More:

https://www.deepmind.com/blog/robocat-a-self-improving-robotic-agent?utm_source=twitter&utm_medium=social&utm_campaign=robocattools

Abstract Background

There has been a remarkable amount of progress in AI in the medical field.

What if you could take a picture of your eye and a computer could tell you if you're likely to develop a heart condition in the next few years based on that picture? This is what this Google AI Model does.

Our eyes have a lot of blood vessels that can provide doctors with information about our health.

Nearly 300,000 eye scans and medical records were used to train Google's model.

The program trained by the model can detect future heart problems.

The process is simple: take a picture of the eye, send it to the AI, and pass the health check.

There are many other problems that can be solved with this capability, including skin problems and detection of genetic diseases.

Read More:

https://beamstart.com/news/googles-ai-revolution-predicting-heart-16871608928

bottom of page