Tesla RoboTaxi - Coming Soon To A Road Near You ?

Hi there.

Another week that seems to be dominated by Elon Musk. He was everywhere. First he was complaining about the crazy talent war between AI companies. Few weeks ago we showed you that yearly salaries for AI engineers were averaging 800.000$ per year. The demand is certainly bigger than the supply.

Then, Musk dropped a bombshell in announcing the Tesla Robotaxi. The official unveiling is on August 8th but he wants to disrupt the automobile industry and claims this car - which has no pedals or steering wheel - is the future of mobility.

The idea is you tell an app where you want to go and your car drives YOU instead of vice versa.

Reactions have been quite lacklustre. Musk has the reputation of underdelivering on his claims.

Let’s see what happens.

Welcome to the Blacklynx Brief

AI NEWS
Update and Upgrade Bonanza

• Elon Musk has raised salaries for Tesla's AI engineers to retain talent amidst intense recruitment battles, notably from OpenAI, describing it as the most severe talent war he has experienced. Musk also noted that Tesla has lost engineers to his other venture, xAI, as well as to aggressive recruiting efforts from competitors like Meta, led by Mark Zuckerberg. This escalation in competition for AI expertise is intensifying the rivalry between Musk and OpenAI, highlighting personal and strategic conflicts in the tech industry.

• Google is contemplating the introduction of a subscription service for advanced AI-driven search features, a first for the tech giant's primary search engine offerings. This proposed premium service would complement Google's existing free search functions but offer enhanced AI capabilities to paying subscribers. The move reflects Google's need to balance the high costs of AI technologies with its lucrative $175 billion search advertising revenue, amid competition from rivals like Microsoft who have already incorporated free AI features into their search tools.

• OpenAI CEO Sam Altman and former Apple designer Jony Ive are aiming to raise up to $1 billion for a new AI-focused device company, leveraging their expertise in AI and design. The device will incorporate OpenAI's conversational AI technology but will not resemble a traditional phone. High-profile investors, including Emerson Collective and Thrive Capital, are potential backers.

• Elon Musk announced that Tesla will unveil its highly anticipated fully autonomous Robotaxi on August 8th, countering rumors that Tesla is scrapping its plans for a more affordable electric vehicle line. This announcement came after Musk refuted a Reuters report claiming Tesla abandoned its entry-level EV project. The Robotaxi, initially promised for 2020 with no pedals or steering wheel, signifies Tesla's ambition in transforming transportation through AI, despite past delays and ongoing legal scrutiny.

• A new study from Cornell University has demonstrated that the performance of AI systems can be significantly enhanced by increasing the number of AI agents working collaboratively on a problem. By having multiple agents independently solve a task and then aggregating their solutions through a voting mechanism, the accuracy of results improves. Remarkably, using this method, a smaller AI model with 13 billion parameters outperformed a larger 70 billion parameter model on certain tasks. This finding suggests that multiplying AI agents could be a cost-effective strategy to boost AI capabilities without necessarily increasing individual model sizes.

• A Reuters report has highlighted a competitive push by major tech companies like Google, Meta, OpenAI, and Apple to acquire extensive online data to enhance their AI models. Following the introduction of ChatGPT, these companies forged a multimillion-dollar agreement with Shutterstock for a vast array of digital content to train their AIs. Additionally, they are investing in private archives, such as Photobucket's, to gain access to billions of historical internet images and videos. This shift from freely scraping the web to purchasing data comes amid increasing legal and ethical challenges surrounding privacy and content ownership.
  

• President Joe Biden announced a significant partnership with Taiwanese chipmaker TSMC, involving the construction of three advanced semiconductor factories in Arizona. This deal, supported by $11 billion in grants and loans, represents a total investment of $65 billion by TSMC and is expected to create over 25,000 jobs. With an additional $50 million allocated for workforce development, the U.S. aims to bolster its domestic semiconductor production to 20% of global output by 2030

• JPMorgan Chase CEO Jamie Dimon emphasized the transformative impact of artificial intelligence in his latest annual letter to shareholders, likening AI's potential to historic technological breakthroughs such as the printing press and electricity. Dimon highlighted that AI could enhance nearly every job at the bank, noting that JPMorgan has already employed over 2,000 AI experts and is actively using AI in 400 different scenarios. His strong endorsement underlines the significant role AI is expected to play in reshaping industries and society, signaling profound changes ahead in how businesses operate.

• Meta has announced the impending release of Llama 3, its advanced large language model, with the first version set to launch within the next month. This release will be followed by multiple versions throughout the year, each with varying capabilities. Although the specific details about the model's size weren't disclosed, it is anticipated to contain approximately 140 billion parameters, positioning it as a competitor to OpenAI's GPT-4

• Intel has launched its Gaudi 3 AI chip, positioning it as a formidable competitor to Nvidia’s leading products. The new chip reportedly outperforms Nvidia's H100 in both training and inference speeds by 50%, and offers 40% greater power efficiency at a lower cost. Set for release in the second quarter of 2024, Gaudi 3 is expected to compete with Nvidia's upcoming H200 series. Alongside this, Intel has expanded its AI offerings with a new enterprise platform and additional Xeon processors.

• Meta also unveiled next generation of its Meta Training and Inference Accelerator AI Chip

• French AI startup Mistral has introduced Mixtral 8×22B, a new large language model (LLM) with significant advancements, which they released as a 281GB file on the platform X for public download. The model, which boasts 176 billion parameters and a 65,000-token context window, utilizes a sparse mixture of experts (SMoE) technique to enhance performance efficiency and reduce costs. This release places Mistral among a flurry of companies, including industry giants like OpenAI, Google, and Meta, that are updating their AI models this week

Closing Thoughts

That’s it for us this week.

I wouldn’t call it a “slow” week, but it was a tad less mind-blowing than usual. We DID discover that upgrades to existing large language models are coming hard and fast.

And giant wads of cash are being invested in the chip industry - because competition in the AI field also manifests in the hardware space - not only in the development of different models

If you want a version of this newsletter with pictures etc. Be sure to subscribe below:

https://www.blacklynxbrief.com/subscribe