Lively Impact Technology Limited

04/05/2022

No more "coding" in the future? 🤔

is a skill that requires specific training and is becoming an essential skill for everyone. In recent years, programming training are included in school curricula, even in primary schools. Instead of teaching everyone the skill, can we do the opposite? Some people propose the concept of -driven . 🔊

There are two parts to voice-driven programming- conversation and proposing programming solutions. The system allows people to simply describe their challenges in plain-spoken language. Then, the AI will produce the code based on the natural language instructions. In other words, users can communicate with the AI system directly to solve programming problems, regardless of whether the user has a background of coding knowledge. Voice-driven programming is a new opportunity for AI to have a more direct impact on customers.

Learn more about this idea 👉

As we start to see AI advance in business, the ways in which we interact with machines are beginning to change. Companies like Salesforce are looking for new opportunities for AI to have a more direct impact on customers. While using AI to surface the customer most likely to churn or most likely to....

27/04/2022

AI model that can do 🤖

Machines learn differently from humans. Humans understand the world with multimodality, combining information in images, sounds, and other details. A research team explored building machines that learn from , and together.

This semi-supervised training method enables out-of-the-box prediction. The AI model can understand the relationship between different elements that change over time. How? The researchers train the AI with millions of videos where snippets of text and audio are masked. Then, the model can learn by matching the correct masked-out snippets. This AI model enables "learning from reentry" prediction, where time-locked correlations enable one modality to educate others, creating commonsense reasoning. 💡

Continue in this article 👉

Most of us have watched television with the sound turned off at one time or another. While it’s usually possible to follow the story at least to some degree, the absence of an audio track tends to…

20/04/2022

⚠️ Google does not accept AI-generated content ⚠️

AI writing tools are powerful in . However, Google's Advocate claimed that AI-generated content is considered spam according to the guidelines. Regardless of how you are using it, you violent the policy. In a serious case, it could lead to a manual penalty.

Actually, Google's Search Engine still lacks the ability to detect auto-generated content. Although it is not happening, we can ask a question- is it a technical degeneration? AI tools help us to perform better, is it a wise decision to ban their use and return to manual operation? 🤔

Learn more about the issue 👉

Content generated automatically using AI writing tools is against Google’s Webmaster Guidelines, says Search Advocate John Mueller.

13/04/2022

A machine learning model that can for another model 🤖

High-quality data is critical for many machines learning tasks. However, choosing a flawless, suitable dataset can be a difficult task. Real data may also touch some privacy and usage rights problems. To add, real data cannot be used to learn about corner cases...

Some MIT researchers develop a to overcome the above limitations. A trained model can generate synthetic data that is indistinguishable from the original data. The researchers reported that their model that is trained with synthetic performs no worse, or even better than models trained using real data. Contrary to the satisfying performance of the generative model, the researchers warned that the synthetic data may reveal some information about the source data, causing security concerns.

Learn more about this ML model 👉

MIT Researchers Developed A Machine-Learning Model To Generate Extremely Realistic Synthetic Data That Can Train Another Model For Downstream Vision Tasks

06/04/2022

AI turns still images into a scene, in seconds 📸

is the process to construct a 3D scene from 2D images taken at different angles. This process uses AI to approximate the behaviour of lights in the real world. NeRF, Neural Radiance Fields, is the technology that uses the approach of the above process. NVIDIA has been a huge contributor to this technology. They recently announced the advanced version of NeRF- Instant NeRFs, which cuts the rendering time significantly into .

The training of NeRFs requires a few dozen images taken from multiple positions around the scene. The camera position of each shot also puts effect. The neural network reconstructs the scene by predicting the colour of light radiating in any direction from any point in the space. The technology is initially time-consuming. However, the Instant NeRF reduces the rendering time and preserve the high-quality result... 💡

Watch the demo video 👉

Instant NeRF is a neural rendering model that learns a high-res 3D scene in seconds — and can render images of it in a few milliseconds.

30/03/2022

Despite the pandemic, the AI market is constantly growing in 2021, we can even see new records in some areas. 2021 sees a record number of AI unicorn births, with an increase of 442% from 2020 to 2021. 🦄

A unicorn is a startup company with a value of over $1 billion. Globally, US and Asia own most AI unicorns. The top 5 unicorn births in 2021 are all from the US, they are Deliverr, SWORD Health, H2O. ai, Paradox and Anyscale.

Subscribe to us now http://www.livelyimpact.com/subscription.html so you won't miss any latest MarTech trends.

23/03/2022

How was the Global AI market last year? 🤔

Despite the ongoing pandemic, Q4'21 is the second biggest for AI funding- the Global AI funding was up 108% YoY in 2021. The Global AI Market was energetic as always and 18% of the total funding was made up by healthcare AI. 👩‍⚕️👨‍⚕️

Subscribe to us now http://www.livelyimpact.com/subscription.html so you won't miss any latest MarTech trends.

16/03/2022

A can resolve more than half of your customer inquiries

"The spiral of misery" describes the situation that the Chatbots are not helpful to customers because the Chatbots do not understand customers' inquiries and give formulaic replies. Eventually, the inquiries are passed to the CS agents. Then, what is the point to invest in a Chatbot? 🤔

Actually, Chatbots can understand questions, but only in fixed phrases. That is because Chatbots are often programmed with a predetermined set of questions and answers. To make the Chatbots responds correctly with different phrases, engineers are utilising AI, namely Conversational AI to improve the accuracy with lesser amounts of training data.

See how Chatbots are evolved with use cases 👉

Customer service chatbots may finally become more intelligent, more conversational and more helpful.

09/03/2022

Meta's vision on building human-level 🤖

The popular approach for building AI with human-level capabilities is to use more data and enlarge the AI models. However, Meta's Chief AI Scientist Yann LeCun proposes an alternative approach inspired by the common sense knowledge in humans and animals.

Humans and animals can predict future outcomes and fill in the missing information with common sense knowledge. Concerning these unique capabilities, LeCun raises an architecture featuring the predictive world model as the centrepiece to predict real-world events. Here comes the first challenge- how to represent multiple plausible predictions when the real world is not always predictable? LeCun's Joint Embedding Predictive Architecture (JEPA) is the key... 🔑

Continue to learn about LeCun's idea 👉

Meta's Chief AI Scientist Yann LeCun sketches how the ability to learn “world models” – internal models of how the world works – may be the key to building human-level AI.

02/03/2022

🤖 grants you the power to respond to unexpected events confidently

Dataminr is an AI-powered tool for and . It fetches data from more than 200,000 public information sources which include social media platforms such as Twitter, Reddit, IoT sensor data, audio transmissions, and even the deep and dark web.

The tool helps identify the earliest signals of high-impact events and emerging risks. Hence, extra time is saved for the organisations to deliberate and respond fast, minimising the impact of unexpected events on them.

Check this AI-powered event and risk detection tool 👉

Customers rely on Dataminr’s AI platform for early signals of high-impact events and emerging risks, to respond with confidence and manage crises more effectively

23/02/2022

How do Nvidia balance image quality and GPU performance with ?

Initially, the image quality of a game is hugely dependent on the monitor's resolution. Previously, Nvidia upscales the image quality with Dynamic Super Resolution (DSR) technique. DSR renders the game at a higher resolution than the monitor can handle and then downscales to the monitor's native resolution. This technique results in better image quality but sacrifices the GPU's performance.

Based on DSR, Nvidia invented a new technique called Deep Learning Dynamic Super Resolution (DLDSR) using the power of Deep Learning 🤖. DLDSR maintains GPU's performance while achieving a similar outcome as DSR.

See how Nvidia made it 👉

For the games your RTX card makes quick work of

18/02/2022

Will we compete with virtual workers in the future?

China plans to be a world leader in AI by 2030. Pursuing the common goal, different Chinese companies have invested heavily in technologies and established 🤖 virtual employees. The companies humanise the AI, give them names and avatars. China Vanke even named their AI-based debt collector as a top employee with a 91.4 per cent success rate in collecting accounts receivable and overdue loan repayments.

Some said these virtual employees set examples with their high performance, other employees will also be encouraged to improve their skills. However, human workers can never achieve the same capabilities and processing power as AI. Here comes a question- will we be under unfair competition in the future? 🤔

Read more about virtual workers in China 👉

China, which plans to be a world leader in AI by 2030, is predicted to deploy more ‘virtual’ employees this year.