The week in AI: Tesla shares surge amid Musk's AI rebranding - even as EV sales plunge

In partnership with

Welcome to The Dispatch! We are the newsletter that keeps you informed about AI. Each Thursday, we aggregate the major developments in artificial intelligence; we pass along the news, useful resources, tools and services, and highlight the top research in the field as well as exciting developments in open source. Even if you aren’t an engineer, we’ll keep you in touch with what’s going on in AI.

NEWS & OPINION

Elon Musk says Tesla should be considered an AI company, not an automaker, as EV sales falter

-------------------------

The world’s most valuable car company had its earnings call for the first quarter of 2024, and announced that its net income had dropped 55 percent from a year ago. It has been an especially disastrous month for Tesla. Just this April, Tesla has announced its first drop in sales since 2020, recalled one line of vehicles, reportedly canceled plans for another, and has begun mass layoffs.

None of those things appear to be of much concern to Mr. Musk. He stated that Tesla should now be thought of as an AI or robotics company, not an automaker, while providing new updates on Tesla’s Optimus autonomous robots. Musk said he believes Optimus will be ‘more valuable than everything else [in the company] combined’.

He revealed that Optimus should be doing ‘useful tasks’ in Tesla factories by the end of the year and sold externally by the end of 2025. While many still look at Tesla and think of cars, Musk’s AI and robotic visions are operating on a grander scale. A new graphic featured Optimus at the top of Tesla’s ecosystem chart, with the robots accounting for ‘a majority’ of the company’s long-term value.

Tesla’s shares surged following the call despite missing earnings expectations.

Zuckerberg: it will take Meta years to make money from generative AI

-------------------------

ChatGPT kicked off the AI chatbot race. Meta hopes to win it - even if it takes a while. It was a major week for Meta in AI: their new AI assistant, powered by Llama 3 (also released this week - see below in the technical section), is now being integrated into the search box of Instagram, Facebook, WhatsApp, and Messenger. It’s also going to start appearing directly in the main Facebook feed.

While it’s full speed ahead in AI for Meta, CEO Mark Zuckerberg asked for patience in the company’s Q1 earnings call. Meta lost 10% of its employees last year, has a weak forecast for Q2, and is planning to spend billions on AI. Their stock is down 15%.

It seems clear Zuck is planning to focus on the adoption of Meta AI first, and revenue later. “There are several ways to build a massive business here, including scaling business messaging, introducing ads or paid content into AI interactions, and enabling people to pay to use bigger AI models and access more compute,” he said. “And on top of those, AI is already helping us improve app engagement, which naturally leads to seeing more ads and improving ads directly to deliver more value.”

Promising AI search startup Perplexity launches Enterprise Pro

-------------------------

Perplexity, the AI search startup that wants to take on Google, has been on a roll of late. It just raised a huge round (with a list of investors including Jeff Bezos, Nvidia, Stanley Druckenmiller and OpenAI cofounder Andrej Karpathy), doubling the startup’s valuation to over $1 billion. They’re now offering an enterprise option for companies after testing it with the likes of Stripe, Zoom, and Databricks who all have good things to say about the service - the blog post goes into detail about enterprise use cases.

Often, enterprise AI is confined to thinking about internal search. Perplexity is betting that external searches are equally (or more) worthy to the company. The new funds raised will go towards the global expansion of the promising startup. It started the chain with SK Telecom from South Korea earlier this year and has added Japan’s SoftBank Corp. and Germany’s Deutsche Telekom to this list.

Search quality has been steadily eroding over time, in part due to generative AI. Perplexity offers a number of features to help improve search. With “Copilot,” a user can narrow down a query by asking clarifying questions. If you ask for ideas on where to host a birthday party for a 2-year-old, for example, Copilot might ask whether you want suggestions for outdoor spaces, indoor spaces or both. If you select “indoor,” it will ask for a rough budget for the party. Only then will it offer a list of possible venues. Perplexity also allows users to search within a specific set of sources, such as academic papers, YouTube videos or Reddit posts. It’s also refreshingly good at admitting when it doesn’t know something.

• Moderna and OpenAI partner to accelerate the development of life-saving treatments

• Meta’s new AI assistant is fun to use, but it can’t be trusted

• Innovation through prompting: democratizing educational technology and more

• Google is combining its Android and hardware teams - and it’s all about AI

• AI starts to sift through String Theory’s near-endless possibilities

• Battle of the bots: getting hired is hard work in the age of AI

• Forget the AI doom and hype, let's make computers useful

• Amazon wants to host companies’ custom generative AI models

• Google consolidates its DeepMind and Research teams amid AI push

• Los Angeles is using AI in a pilot program to try to predict homelessness and allocate aid

Boost Your Marketing Performance with Anyword

Even the largest AI models don’t know what works for your marketing. They don’t know your brand, audience, or what resonates. Anyword does.

Trusted by over 1M+ marketers, Anyword generates optimized content trained on your marketing channels, with predictive performance scoring & insights for any copy, channel, and audience – so you don’t have to guess what content will perform best.

Easily create engaging, on-brand content at scale that boosts marketing performance and achieves team goals.

Try Anyword now!

TECHNICAL, RESEARCH & OPEN SOURCE

Meta’s Llama 3: the most capable openly available LLM to date

-------------------------

Meta has finally unveiled its next generation of open source Llama 3 models, starting with 8B and 70B parameter variants. These models will soon be available on Hugging Face and other platforms, and currently you can chat with Llama 3 (for free) on Meta AI.

The models are the best open source models so far for their respective sizes. Llama 3 8B outperforms all other similar-sized open source models like Gemma 7B and Mistral 7B Instruct across all benchmarks by a fair margin. Llama 3 70B scored 79.5 in MMLU, which is the highest MMLU score for any open source model yet - also outperforming Gemini 1.5 Pro and Claude-3 Sonnet.

The Llama 3 family’s largest model is still in training and will have over 400B parameters, a larger context window, be multimodal, and is expected to be on par with GPT-4 and Claude 3 Opus. It’s unclear when (or even if!) this biggest model will be open sourced.

Microsoft launches Phi-3: redefining what’s possible with small language models (SLMs)

-------------------------

While large language models (LLMs) will still be the gold standard for solving many types of complex tasks, Microsoft has been developing a series of small language models (SLMs) that offer many of the same capabilities found in LLMs but are smaller in size and are trained on smaller amounts of data.

The company announced the Phi-3 family of open models, the most capable and cost-effective small language models available. Microsoft is making the first in that family of more powerful small language models publicly available: Phi-3-mini, measuring 3.8 billion parameters, which performs better than models twice its size - and can run locally on your phone.

It’s available in the Microsoft Azure AI Model Catalog and on Hugging Face, as well as Ollama, a lightweight framework for running models on a local machine. Microsoft also announced additional models to the Phi-3 family are coming soon to offer more choice across quality and cost.

MORE IN T/R/OS

• Large language models approach expert-level clinical knowledge and reasoning in ophthalmology: A head-to-head cross-sectional study

• Snowflake’s Arctic: the best LLM for enterprise AI - efficiently intelligent, truly open

• The Open Medical-LLM Leaderboard: Benchmarking Large Language Models in Healthcare

• Apple releases OpenELM: small, open source AI models designed to run on-device

• Apple reportedly developing its own custom silicon for AI servers

That’s all for this week! We’ll see you next Thursday.’