Despite export limitations put in place by the United States, Chinese sellers are profiting from the soaring demand for high-end Nvidia chips, particularly the A100 artificial intelligence (AI) chips. People looking for the sought-after chips have started gathering in Shenzhen’s well-known Huaqiangbei electronics district, which is well-known for its extensive selection of electronic products, including drones and camera parts.
Although the chips are not advertised in public, they can be found with careful research. The A100 chips, created by US chip company Nvidia, are reportedly offered for a stunning $20,000 per unit, which is double the standard price.
High-end US chips can be purchased or sold in China legally, but because of US export restrictions, a black market has emerged. Vendors take care to avoid drawing the attention of Chinese or US authorities.
In an effort to halt Chinese advancements in artificial intelligence (AI) and supercomputing in the face of escalating political and trade tensions, President Joe Biden’s administration introduced export restrictions. Thought to be the best for machine learning tasks, Nvidia’s microprocessors are now in high demand because of the worldwide AI boom, which was spurred by the popularity of OpenAI’s ChatGPT.
The precise number of A100 and H100 chips entering China is still unknown, but Reuters discovered that several dealers in Hong Kong and mainland China have easy access to small amounts of A100 chips. Buyers often include entrepreneurs, gamers, researchers, and app makers. Notably, a few municipal governments in China are also among the buyers.
In reaction to the circumstance, Nvidia declared that they do not allow the export of the A100 or H100 to China and instead provide lower-capability alternatives that adhere to US laws. The business also stated that it will punish consumers who were discovered to be breaking the terms of their agreement.
A radio station in the US made history by hiring the first full-time DJ powered by artificial intelligence in a ground-breaking move. For their midday broadcast, KBFF Live 95.5 FM, owned by Alpha Media, has used Futuri Media’s RadioGPT software to develop an AI version of their host Ashley Elzinga.
We made history as the world's first radio station with an AI DJ! Our midday host Ashley has become AI Ashley! We can’t wait for you to meet Ashley, the world's first artificially intelligent DJ. As to the intelligence of our other DJ’s…we’ll save that for another post 😉 pic.twitter.com/CtlMhYU0IO
The creative project aims to improve the station’s capacity for content creation and provide audiences and clients with more timely and thorough information.
Phil Becker, EVP of Content at Alpha Media, expressed his excitement about the possibilities of AI in radio broadcasting and claimed that it allows the station to be more adaptable than ever. They are now able to feature their content creators more frequently and in more situations, thanks to the deployment of RadioGPT.
Elzinga may be seen making a joke about the AI host and hinting that she is off work in a recent video that the radio station posted on Twitter. To allay worries among regional DJs, Alpha Media emphasizes that the AI DJ will not take the place of the actual Ashley Elzinga. While working with her AI counterpart, Elzinga will continue to be paid as usual.
In the video, the AI presenter, who was inspired by Elzinga, manages to replicate her voice very well, sounding kind and relaxed when speaking with viewers. A lucky caller is told in the video they have won tickets to a Taylor Swift concert in one video by the AI Ashley.
According to the 2023 Hong Kong Salary Guide published, artificial intelligence (AI) technology would eliminate almost 800,000 jobs by 2028, or 25% of the workforce in Hong Kong. The report names customer service agents, administrative workers, and data entry clerks as the most vulnerable job categories to AI impact.
Beyond these industries, the impact of AI is anticipated to affect careers including law, translation, illustration, and content creation. Concerns about significant employment losses have been raised by the rising popularity of AI models like ChatGPT. According to sources, several organizations in Hong Kong are now encouraging people in roles that previously did not require IT skills to become proficient in ChatGPT.
According to a study by the major investment bank Goldman Sachs, AI has the potential to automate 25% of all vocations, with administrative jobs being the most automatable at 46%, followed by legal jobs at 44%, and professionals in architecture and engineering at 37%.
Researchers from Nanyang Technological University in Singapore and the Damo Academy of Alibaba Group discovered that employing large language models (LLM) for data analysis, such as GPT-4 can be done for less than 1% of the price of hiring a human analyst, while providing equivalent results.
The report emphasizes the possible risk to employment security posed by the growing use of generative AI. Additionally, GPT-4 completes jobs substantially faster than people do, and in some circumstances, it outperforms human data analysts in terms of accuracy and analysis.
The text-generating AI models GPT-3.5-turbo and GPT-4 have been updated by OpenAI to include a feature called function calling. Developers can now specify functions for GPT-4-0613 and GPT-3.5-turbo-0613, and the model will intelligently decide which JSON object to output that has the arguments needed to invoke those functions.
This is a novel approach to more securely link GPT’s capabilities with external tools and APIs, according to a blog post published by OpenAI on Tuesday. Developers can build chatbots that respond to queries by calling external tools (such as ChatGPT Plugins) using the function calling feature.
For instance, it can convert queries such as “What’s the weather like in Boston?” to get_current_weather(location: string, unit: ‘celsius’ – ‘fahrenheit’) or “Email Anya to see if she wants to get coffee next Friday” to a function call like send_email(to: string, body: string). OpenAI indicated that it will start the process of upgrading and deprecating the initial releases of GPT-4 and GPT-3.5-turbo, which it announced in March.
On June 27, all software utilizing the stable model names (GPT-3.5-turbo, GPT-4, and GPT-4-32k) will be automatically updated to the new models. Additionally, the business announced a 25% price reduction for the GPT-3.5-turbo.
Developers can use GPT-3.5-turbo for $0.0015 per 1,000 input tokens and $0.002 per 1,000 output tokens, yielding about 700 pages per dollar. A pricing reduction is also being applied to Text-embedding-ada-002, one of OpenAI’s most popular text embedding models.
Major changes have been revealed by The Recording Academy for the biggest award show in music. The Grammys will now accept entries for music created by artificial intelligence.
However, there is a catch. According to the New York Post, AI-generated music must also include meaningful and relevant human elements. Only human creators are qualified to be nominated for or awarded a Grammy Award, according to the most recent regulations.
According to the guidelines, AI-based content without human authorship is not eligible for the competition, while AI-based content with human authorship is. The latest adjustments coincide with an increase in deep-fake tracks and AI-generated music. This year, David Guetta surprised his audience by performing an Eminem song, even though the latter had never recorded it.
Recently, Universal Music Group has removed a song with AI-generated vocals from streaming sites that falsely claimed to be from Drake and the Weeknd after it went viral. The song, Heart on My Sleeve, was denounced by the record company for “infringing content created with generative AI“.
Spotify, the biggest player in the audio streaming industry, recently removed tens of thousands of songs, or roughly 7% of the recordings that the platform Boomy had uploaded. According to an individual familiar with the incident, the major record label Universal Music had alerted all of the major streaming services of unusual streaming activity on Boomy recordings.
Meta has announced a ground-breaking generative AI model called “Voicebox” that has the power to transform speech synthesis. Voicebox, according to a blog post by Meta, is the first model that can perform well for speech-generation tasks, even without specialized training for such tasks.
Voicebox is an expert in creating high-quality audio snippets, unlike conventional models that generate visuals or text. It has the ability to create speech in a variety of styles, either from scratch or by adjusting existing samples. Six languages, including German, Spanish, English, French, Polish, and Portuguese are supported by the model for speech synthesis. Voicebox also provides functions including noise reduction, content editing, style conversion, and varied sample production.
Voicebox is distinguished by its distinctive learning methodology. Voicebox learns directly from the untranscribed audio and the corresponding transcriptions rather than using autoregressive models. As a result, the model is more flexible and versatile because it can alter any portion of a given sample.
According to Meta, when given the surrounding speech and its associated transcript, Voicebox can be trained to predict a speech segment. Once the model has mastered the capacity to complete speech depending on context, it can be used for a variety of speech production tasks, enabling it to produce only the necessary parts of an audio recording rather than the entire recording.
Voicebox excels in a variety of applications because of its adaptability, such as in-context text-to-speech synthesis, cross-lingual style transfer, voice denoising and editing, and diversified speech sampling. Performance and adaptability of the model open up new avenues for creative audio generation and advanced speech manipulation.
Google unveiled a virtual try-on function earlier this week, powered by generative AI. The new function displays how clothing appears on a variety of body types. Google claims that the new function would enable customers to tweak products until they discover the ideal one.
Machine learning and new visual matching algorithms that enable users to fine-tune inputs like color, pattern, and style make this possible. Users will be able to see selections from stores all across the internet, which is the main benefit in this case.
Customers can now view how a piece of clothing will drape, stretch, cling, fold, and create wrinkles and shadows on a variety of models, thanks to the virtual try-on. Additionally, the tool will assist customers in locating complementary apparel items in various sizes, styles, or colors.
Google, in the company’s blog post, said, “We chose persons ranging in sizes XXS-4XL, representing varied skin tones (using the Monk Skin Tone Scale as a guide), body forms, ethnicities, and hair types.”
Google used multiple sets of photos of more than 80 models in a variety of stances, representing a range of sizes, skin tones, body shapes, and ethnicities, to create the virtual try-on function. In order to produce accurate photographs of the subject from all angles, the AI-powered tool learned from all the models how to match the shape of certain apparel in various postures.
According to Google, the feature would initially be compatible only for women’s clothing with retailers like H&M, Everlane, Anthropology, etc. It will eventually diversify into men’s clothes, mostly shirts. Over time, it’s anticipated that the tool will grow more exact and accurate.
This week, Google introduced a Lens image search function called DermAssist that can identify skin conditions such as odd mole or rash. Users can get more information concerning a bump on the lip, a line on a nail, or hair loss from the scalp. The tool functions on all areas of the body.
“Just take a picture or upload a photo through Lens, and you’ll find visual matches to inform your search,” the blog post says. Importantly, Google makes it clear that the material provided is informational only and not a diagnosis. The blog advises users to consult medical professionals for accurate diagnosis.
According to Google spokesperson Craig Ewer, the feature is accessible to everyone in the US and supports all major languages. For years, Google has been investigating the application of AI picture identification to skin diseases.
A project that aimed to diagnose skin, hair, and nail issues using a combination of photographs and survey responses was showcased by Google during its I/O developer conference in 2021. At the time, Google claimed that the technology could distinguish between 288 different conditions and would, 84% of the time, present the proper condition in the top three possibilities.
According to Google’s DermAssist tool page, it is “currently undergoing further market testing through a limited release.” It also states that despite being CE-marked as a Class 1 Medical Device (a label for products in the European Economic Area), it has not undergone FDA review.
Applications are being accepted for a free online course called “Introduction to Machine Learning” through the NPTEL platform, offered by IIT Kharagpur. The course provides a brief review of machine learning‘s core ideas as well as some of the most popular algorithms in the industry.
Students will gain knowledge of the most well-known supervised learning algorithms, such as the naive Bayes algorithm, logistic regression, decision trees, and k-nearest neighbors. They will also be given an introduction to Bayesian learning. In addition, the course will include some tutorial sessions and hands-on Python programming problem-solving.
Sudeshna Sarkar, a professor and the head of the IIT Kharagpur Department of Computer Science and Engineering, will teach the course. She earned her BTech in 1989 from the IIT Kharagpur, her MS in 1995 from the University of California, Berkeley, and her PhD in 1995 from the same institution. Although the course is free, students have to pay an exam fee of ₹1000 to get a certificate of completion and two credit points.
The application of machine learning algorithms, hypothesis space, overfitting, bias, and variance, tradeoffs between representational power and learnability, evaluation methods, and cross-validation are just a few of the topics that students will gain insight into.
The course will start on July 24, 2023, and end on September 15, 2023. The enrollment deadline is July 31, 2023. The registration date for the exam is August 18, 2023. The test will take place on September 24, 2023, at IST (Indian Standard Time).
The rise of remote work has revolutionized the way we work, allowing teams to collaborate across different time zones and geographical boundaries. While remote work offers numerous advantages, it also presents unique challenges, such as the potential for isolation and lack of team cohesion. This is where virtual team-building activities come into action.
These activities not only foster strong relationships but also enhance communication, trust, and collaboration within remote teams. In this post, we’ll explore effective virtual team building activities to strengthen your remote team and create a more connected and engaged workforce.
Understanding Virtual Team Building
Virtual team building refers to the intentional efforts made to build camaraderie, trust, and a sense of belonging within remote teams. It involves a range of activities designed to promote interaction and collaboration, despite the physical distance between team members. The benefits of virtual team building are manifold, including increased morale, improved communication, enhanced productivity, and a stronger sense of team identity.
However, it’s essential to acknowledge the challenges that come with virtual team building, such as time zone differences, varying technological proficiency, and potential feelings of disconnect. By approaching these challenges with thoughtful preparation and effective execution, it’s possible to overcome them successfully.
Team Building Activities for Remote Teams
Here are some of the activities that you can embrace.
Icebreaker Activities to Establish Rapport
Icebreaker activities are a fantastic way to kickstart team-building efforts and create a comfortable and friendly atmosphere. Encourage team members to introduce themselves and share fun facts about their lives. This helps break the ice and initiates conversations beyond work-related topics.
Another popular icebreaker game is “Two Truths and a Lie,” where team members share three statements about themselves, and others have to guess which one is false. Additionally, organizing “Show and Tell” sessions allows team members to showcase something meaningful to them, fostering deeper connections and understanding.
Collaborative Problem-Solving Activities
Collaborative problem-solving activities promote teamwork, critical thinking, and creative problem-solving skills within remote teams. Virtual scavenger hunts are a great option, where teams work together to find specific items or solve riddles using online resources.
Online escape rooms provide an immersive experience, requiring team members to work together to solve puzzles and escape within a time limit. Team-based trivia or quizzes encourage friendly competition while improving knowledge and collaboration.
Communication and Trust-Building Activities
Communication and trust are the foundations of successful team dynamics. Virtual team-building games like “Never Have I Ever” can be adapted for remote settings, allowing team members to share personal experiences and anecdotes.
The “Storytelling Circle” exercise involves passing a story from one team member to another, stimulating creativity and collaboration. Virtual coffee or lunch breaks provide casual spaces for team members to connect, chat, and build relationships beyond work-related matters.
Creative Activities to Encourage Innovation
Encouraging creativity and innovation is crucial for remote teams. Virtual brainstorming sessions allow team members to collectively generate ideas and solutions for specific challenges or projects.
Online art or music challenges can tap into the creative talents of team members, fostering a sense of pride and accomplishment. Virtual team-building workshops or seminars led by experts can also provide learning opportunities while promoting collaboration and out-of-the-box thinking.
Health and Wellness Activities
Taking care of the well-being of remote team members is essential for their overall satisfaction and productivity. Virtual fitness challenges, such as step challenges or yoga sessions, promote physical well-being while encouraging friendly competition.
Mindfulness or meditation sessions can help reduce stress and promote a positive mindset within the team. Virtual cooking or recipe-sharing activities enable team members to share their culinary skills and bond over a shared love for food.
Key Considerations for Successful Implementation of Virtual Team Building Activities
While incorporating virtual team-building activities, there are a few best practices to keep in mind. Firstly, plan and schedule activities in advance, considering the time zones and availability of team members. Ensure that the activities are inclusive and accommodating to everyone’s schedules.
Next, choose the right tools and platforms to facilitate seamless virtual team building. Video conferencing platforms with breakout rooms, interactive whiteboards, and chat features can enhance engagement and collaboration.
Make sure to create an inclusive and engaging environment by providing clear instructions, setting expectations, and encouraging active participation from all team members. Consider incorporating a mix of synchronous and asynchronous activities to accommodate different working styles and time zones.
Lastly, collect feedback from team members to gauge their preferences and adapt the activities accordingly. Regularly assess the impact of virtual team-building efforts and make necessary adjustments to meet the evolving needs of the team.
Final Thoughts
Since there are projected to be 36.2 million American employees working remotely by 2025, virtual team-building activities are invaluable for creating a strong and connected remote team. By implementing long-term strategies and prioritizing team cohesion, remote teams can thrive and achieve success together, regardless of physical distance.
