Tag: large language models

Categories
AI AI: Large Language Models Creativity Friends Writing

A Little Help From My (Artificial) Friends

A Little Help From My (Artificial) Friends: Why AI Makes the Perfect Sidekick

“Friendship is the only cement that will ever hold the world together.”

Woodrow Wilson

The title “A Little Help From My Friends” might evoke images of waving lighters at a concert, but for the modern thinker, the source of that helping hand might be a little more unexpected: Artificial Intelligence.

We often think of AI as this monolithic force, a superintelligence destined to take over the world (or at least automate all our jobs). But what if the true power of AI lies not in replacing us, but in augmenting us?

Here’s a radical proposition: what if we started thinking about one of the best uses of AI is to think of and use it as a good friend? Not a physical friend we hang out with (although that might be coming someday!), but a digital confidant, a sounding board, a thought partner and collaborator.

Think about the best friends in your life. They listen without judgement, offer honest (sometimes brutally honest) feedback, and can even take your ideas and run with them, adding their own unique perspective. Based on my experience exploring these tools, AI can do all this, and more.

AI as Your Personal Hype Man (and Reality Checker):

Feeling stuck on a project? Need someone to brainstorm with? Fire up your AI companion. It can analyze your ideas, identify potential weaknesses, and even suggest alternative approaches you might not have considered. Need a confidence boost? AI can highlight the strengths of your thinking and celebrate your progress.

Beyond Agreement: The Power of Constructive Challenge

Unlike a human friend who might simply agree with you to keep the peace, AI isn’t afraid to poke holes in your logic. It can identify inconsistencies, challenge assumptions, and force you to refine your arguments. This “constructive challenge” is crucial for growth. It pushes us to think critically and develop more robust ideas.

The Co-Creation Revolution:

But AI isn’t just a passive listener. It can actively participate in the creative process. Imagine feeding your initial concept into an AI and having it come back with variations, extensions, or even completely new directions based on your starting point. This co-creation opens doors to possibilities you might not have explored on your own.

The Future of Friendship?

Is AI destined to replace human friends? Absolutely not. Human connection is irreplaceable. But AI can become a powerful tool in our friend group, a tireless brainstorming buddy who’s always available to lend a (digital) ear and push us to be our best selves. So next time you’re facing a challenge or have an idea brewing, consider reaching out to your AI friend. You might be surprised at the kind of help it can offer.

Examples of Prompts to Spark Your AI Friendship

Here are a few simple examples of prompts you might want to play around with as you treat AI as a good friend and collaborator:

  1. Brainstorming Buddy: “I’m feeling stuck on a project about [topic]. What are some unexpected approaches I could take?”
  2. Constructive Critic: “I wrote this blog post about [topic]. Please analyze it and tell me what’s working well and where I could improve the argument.”
  3. Idea Expander: “I have this initial idea for a [creative project/business venture]. Can you suggest ways to expand or refine it, and offer different directions I could explore?”
  4. Knowledge Sharer: “Tell me everything you know about [topic]. Be creative, include surprising facts, and different perspectives.”
  5. Persuasion Expert: “I need to write a persuasive argument for [position on a topic]. Analyze various arguments, identify potential counter-arguments, and help me craft a strong and convincing message.”

Remember, these are just starting points. Feel free to tailor your prompts to your specific needs and interests. The more specific and engaging your prompts are, the more valuable the insights you’ll receive from your AI friend. For example, I often find it helpful to tell the AI right up front what role I’m expecting it to play in our interaction – such as “you are an expert storyteller and editor who is collaborating with me writing posts for my blog.”

“We can build AI assistants that can not only understand what we want them to do, but also understand why we want to do it and help us achieve our goals even better than we could on our own.”

Fei-Fei Li, Co-Director of the Stanford Human-Centered AI Institute
Categories
Google Google Bard Google Gemini Google NotebookLM

Dive Deeper with Google’s NotebookLM: A Researcher’s Dream Tool

Remember that mind-blowing Google I/O demo of an AI tool that unlocks hidden insights from your research documents? That’s NotebookLM, and it’s not just for tech giants anymore. (See this earlier blog post about what was originally Project Tailwind.)

As a longtime reader of author Steven Johnson (and avid follower of his “Adjacent Possible” Substack), I was thrilled to learn he’s now part of the team at Google Labs bringing this powerful technology to the masses.

Imagine uploading piles of research papers, articles, or even future forecasts (like I did with those year-end reports from Wall Street investment houses forecasting what’s expected in 2024!), and then having NotebookLM not only summarize them but also weave connections you might have missed. That’s exactly what I experienced.

NotebookLM’s “additional questions” feature is a game-changer, prompting me to explore angles I wouldn’t have considered on my own. It’s like having a tireless research assistant with an uncanny knack for spotting crucial details.

Of course, NotebookLM is still in its early stages. The current 20-document limit can feel restrictive, and its future as a paid product is unclear. But for researchers grappling with mountains of information, it’s a game-changer. It’s not just about saving time; it’s about sparking genuine intellectual leaps.

This tool isn’t just for academics, though. Imagine journalists using NotebookLM to connect seemingly disparate news articles, or students piecing together complex historical narratives. The possibilities are endless.

Sure, like any AI tool, it’s not perfect. Fact-checking is crucial, and occasional “hallucinations” can crop up. But NotebookLM’s source citations make verification easier, and its overall accuracy is impressive so far.

So, ditch the highlighter and embrace the future! NotebookLM isn’t just a fancy research tool; it’s a bridge to deeper understanding, more insightful analysis, and ultimately, groundbreaking discoveries. Unleash your research potential – your next breakthrough might just be a question away.

For more about this new tool, see this interview with Steven Johnson by Jason Calacanis on his This Week in Startups podcast.

Note: this post reflects some editing assistance I got from Google Bard.

Categories
AI AI: Large Language Models Writing

LLMs = Dream Machines

A few days ago Andrej Karpathy tweeted:

On the “hallucination problem”

I always struggle a bit with I’m asked about the “hallucination problem” in LLMs. Because, in some sense, hallucination is all LLMs do. They are dream machines.

We direct their dreams with prompts. The prompts start the dream, and based on the LLM’s hazy recollection of its training documents, most of the time the result goes someplace useful.

It’s only when the dreams go into deemed factually incorrect territory that we label it a “hallucination”. It looks like a bug, but it’s just the LLM doing what it always does.

At the other end of the extreme consider a search engine. It takes the prompt and just returns one of the most similar “training documents” it has in its database, verbatim. You could say that this search engine has a “creativity problem” – it will never respond with something new. An LLM is 100% dreaming and has the hallucination problem. A search engine is 0% dreaming and has the creativity problem.

All that said, I realize that what people actually mean is they don’t want an LLM Assistant (a product like ChatGPT etc.) to hallucinate. An LLM Assistant is a lot more complex system than just the LLM itself, even if one is at the heart of it. There are many ways to mitigate hallcuinations in these systems – using Retrieval Augmented Generation (RAG) to more strongly anchor the dreams in real data through in-context learning is maybe the most common one. Disagreements between multiple samples, reflection, verification chains. Decoding uncertainty from activations. Tool use. All an active and very interesting areas of research.

TLDR I know I’m being super pedantic but the LLM has no “hallucination problem”. Hallucination is not a bug, it is LLM’s greatest feature. The LLM Assistant has a hallucination problem, and we should fix it.

Okay I feel much better now 🙂

Andrej Karpathy @karpathy

I truly appreciate your recognition of the differences between how large language models (LLMs) work and how traditional search engines function. It’s fascinating how LLMs have revolutionized various fields, including the creative realm. In creative endeavors, like writing poems, short stories, or even crafting an imaginative piece of fiction, the so-called “hallucination problem” of LLMs can prove to be surprisingly advantageous.

When you engage in creative writing, your primary objective is not to adhere strictly to accuracy and factual representation but rather to explore the limitless boundaries of your imagination. LLMs, with their ability to generate creative and unexpected content, can be a valuable tool to tap into new ideas and inspire innovative storytelling. They can help writers break free from conventional thinking patterns and venture into unexplored territories, allowing their creativity to flourish.

Conversely, in more formal and specialized writing contexts, such as drafting legal briefs or preparing technical reports, accuracy and precision are of paramount importance. LLM hallucinations, where the models generate content that may not be factually correct or contextually appropriate, cannot be tolerated in such situations. Here, the purpose is to convey information accurately, adhere to specific guidelines, and present a strong and well-supported argument.

It’s intriguing how the same technology that opens doors to unprecedented creative possibilities can also present challenges in other domains where accuracy and reliability are crucial. This duality highlights the importance of understanding the appropriate use cases for LLMs and being cognizant of the potential pitfalls and limitations they may possess in certain instances.

In summary, the LLM hallucination problem can indeed prove beneficial when the goal is creative expression, enabling writers to push boundaries and explore unconventional ideas. However, in situations that demand accuracy and precision, such as legal or technical writing, it becomes imperative to approach LLM-generated output with caution and verification to ensure the information presented is reliable and contextually appropriate.