Illustration of a person reading a book in a giant library
Technical Insights

The Unexpected Power of Large Language Models: Training on massive amounts of text partly offsets lack of exposure to other data types.

Recent successes with large language models have brought to the surface a long-running debate within the AI community: What kinds of information do learning algorithms need in order to gain intelligence?
Illustration of two red and blue toy robots fighting with a yellow background
Technical Insights

Do Large Language Models Threaten Google?: ChatGPT and other large language models could disrupt Google's business, but hurdles stand in the way.

In late December, Google reportedly issued a “code red” to raise the alarm internally to the threat of disruption of its business by large language models like OpenAI’s ChatGPT. Do large language models (LLMs) endanger Google's search engine business?
Andrew Ng on a couch with a cup of coffee and a book
Technical Insights

How to Achieve Your Long-Term Goals: Make your projects add up to achievement by charting a path and gathering advice from mentors.

As we enter the new year, let’s view 2023 not as a single year, but as the first of more in which we will accomplish our long-term goals. Some results take a long time to achieve, and even though...
A red Facebook dislike button surrounded by dozens of Facebook like buttons
Technical Insights

Should AI Moderate Social Media? Deciding which posts to show or hide is a human problem, not a tech problem.

What should be AI’s role in moderating the millions of messages posted on social media every day? The volume of messages means that automation is required. But the question of what is appropriate moderation versus inappropriate censorship lingers.
Question asked by Andrew Ng and answered by the latest version of ChatGPT
Technical Insights

When Models are Confident — and Wrong: Language models like ChatGPT need a way to express degrees of confidence.

One of the dangers of large language models (LLMs) is that they can confidently make assertions that are blatantly false. This raises worries that they will flood the world with misinformation. If they could moderate their degree of confidence appropriately, they would be less likely to mislead.
Screen capture of WhyLabs functioning
Technical Insights

AI, Privacy, and the Cloud: How One Cloud Provider Monitors AI Performance Remotely Without Risking Exposure of Private Data.

On Monday, the European Union fined Meta roughly $275 million for breaking its data privacy law. Even though Meta’s violation was not AI specific, the EU’s response is a reminder that we need to build AI systems that preserve user privacy...
Screen capture of Galactica demo
Technical Insights

What the AI Community Can Learn from the Galactica Incident: Meta released and quickly withdrew a demo of its Galactica language model. Here's what went wrong and how we can avoid It.

Last week, Facebook’s parent company Meta released a demo of Galactica, a large language model trained on 48 million scientific articles. Two days later, amid controversy regarding the model’s potential to generate false or misleading articles, the company withdrew it.
Illustration of Andrew Ng on a computer searching for "Panda bear" and getting a Paddington instead
Technical Insights

Prompt Engineering: Future of AI or Hack?

Is prompt engineering — the art of writing text prompts to get an AI system to generate the output you want — going to be a dominant user interface for AI? With the rise of text generators such as GPT-3 and Jurassic and image generators such as DALL·E...
Left: Panda drawn by Andrew Ng in 10 minutes | Right: Panda generated by Stable Diffusion in seconds
Technical Insights

Text-to-Image Generation and the Path to Truly Open AI

Stable Diffusion, an image generation model that takes a text prompt and produces an image, was released a few weeks ago in a landmark event for AI. While similar programs can be used via API calls or a web user interface, Stable Diffusion can be freely downloaded and run on the user’s hardware.
A statue of Lady Justice holds a set of scales in each hand, signifying inconsistent decision making.
Technical Insights

Toward More Consistent Decision-Making

Andrew Ng considers how inconsistent human decisions are, and how AI can reduce that inconsistency.
Vehicle avoiding traffic cones
Technical Insights

The Trouble With Reinforcement Learning

While working on Course 3 of the Machine Learning Specialization, which covers reinforcement learning, I was reflecting on how reinforcement learning algorithms are still quite finicky.
Microscope and Plato's head statue
Technical Insights

Can an AI System Be Sentient? Ask a Philosopher

A Google Engineer recently announced he believes that a language model is sentient. I’m highly skeptical that any of today’s AI models are sentient. Some reporters, to their credit, also expressed skepticism.
Blue balloon on nails in a light pink background
Technical Insights

How to Build AI Startups Part 3: Set Customer Expectations!

One of the challenges of building an AI startup is setting customer expectations. Machine learning is a highly experiment-driven field. Until you’ve built something, it’s hard to predict how well it will work.
Jupyter Notebooks
Technical Insights

Presenting a Technical Concept? Use a Jupyter Notebook

Machine learning engineers routinely use Jupyter Notebooks for developing and experimenting with code. They’re a regular feature in DeepLearning.AI’s courses. But there’s another use of Jupyter Notebooks that I think is under-appreciated.
Wooden ladder pointing to the sky with a hand on it
Technical Insights

How AI Can Help Achieve Humanity's Grand Challenges

Last week, I wrote about the grand challenge of artificial general intelligence. Other scientific and engineering grand challenges inspire me as well. For example, fusion energy, extended lifespans, and space colonization have massive...

Subscribe to The Batch

Stay updated with weekly AI News and Insights delivered to your inbox