- The AI Plug
- Posts
- #43 Google's AI search is too slow
#43 Google's AI search is too slow
AND: An exclusive interview with Daniel Nguyen
The AI Plug
June 05, 2023
Next time you start getting hangry and snap at your colleague because you haven’t snacked, consider this: underfed jumping spiders literally go blind, with malnutrition decaying their photoreceptors. How does this relate to AI? It doesn’t, but now you’re listening, let’s dive in.
In today’s newsletter:
Top News: Seach Generative Experience is slow, Mostaque’s a liar, and Intel’s new AI model is six times the size of GPT.
TL;DR Rundown: Google Docs has more AI features, what The Matrix can teach us about AI, and GPT4 + Minefield is mind-blowing
Tool of the day: Web content sent directly to your Kindle
7 Questions: Exclusive interview with Daniel Nguyen
The point of it all: What’re your thoughts?
Top News 🔝
Three biggest stories if you’re in a rush
Ask Google anything, like, “When was AI born?,” and the world’s largest search engine takes 0,44 seconds to collate ~362,000,000 results. That’s quicker than your babysitter’s boyfriend when your car pulls up. It’s also what we expect from Google - rapid, if not immediate, results to any query we can think of. So why would we want that to change?
We wouldn’t. But in a bid to stay competitive with Open AI and Microsoft, Google is rolling out a new AI-powered search assistant, Search Generative Experience (SGE). Problem is, it’s slower than a wet week in a thunderstorm. Users with early access to the search feature report sub-par answers, with cluttered results pages and long loading times, sounding more like Aliweb from 1992 rather than a cutting-edge AI-search tool from the world’s 4th largest company in 2023.
Indeed, blunders with their own chatbot Bard back in February cost Google over $120B, or 10% of their market value, really putting Google under the pump to come up with a competitive alternative to ChatGPT and Bing. It’s still early stages, and if past performance is ever an indicator of future success, we can expect Google to improve SGE in the near future. For now, though, it’s still the good old top 3 links in the SERP.
- - - - - - - - - - - - - -
With an Oxford master’s degree, Emad Mostaque, the award-winning hedge-fund manager, advisor to the United Nations, and driving force behind the $1B company Stable Diffusion (SD), “kicked off the AI gold rush” (his words) when SD released their text-to-image generator last summer.
Interestingly enough, those are all his words. He doesn’t have a master’s degree from Oxford, but rather a bachelor’s. His hedge fund collapsed four months after their successful year, and the U.N. don’t speak with him. Employees from Stable Diffusion say:
“He’s great at taking other people’s work and putting his name on it or doing stuff that you can’t check if it’s true.”
Outlets fault Mostaque for “embellishment” and “mischaracterizations,” (to put it lightly). By fabricating strategic partnerships with Amazon (denied) and lying about the OECD, WHO, and World Bank as Stability Diffusion’s partners (all of whom have denied cooperating with Mostaque), one of the leading faces of the AI revolution is, at least partially, a massive fraud. That does seem to come with the territory, though, with publically available AI tech primarily being used to fake images, events, and people, so perhaps it’s fitting.
- - - - - - - - - - - - - -
Intel, a US manufacturer of graphics chips, flash memory, and motherboard chipsets, recently announced at the ISC23 keynote “Aurora gen AI” - a generative AI model with a trillion parameters, which is almost six times more than ChatGPT. Intel is focused on building this model to help the scientific community accelerate “System Biology, Cancer Research, Climate Science, Cosmology, Polymer Chemistry, and Materials Science.”
At the moment, this remains nothing but an announcement. Nonetheless, if Intel delivers here, they’ll have a model theoretically capable of understanding and perceiving sensitive and nuanced topics such as political scenarios and policy-making, which could have far-reaching societal and governmental effects far sooner than anticipated.
TL;DR Rundown 🐂
Summary of note-worthy trending articles
ChatGPT took their jobs; now they’re walking dogs and fixing air conditioners. (link)
Why Shoggoth, an octopus-like sci-fi character, now symbolizes the state of AI. (link)
Can The Matrix and other sci-fi films teach us anything about AI threats? (link)
Plugging GPT4 into Minecraft unearthed an entirely new potential for AI. (link)
A glimpse into what an AI doomsday could look like. (link)
We should treat AI like a biological weapon, not a nuclear one. (link)
Character.ai provides ChatGPT-style conversations in famous people’s voices. (link)
Google Docs, Slides, and Gmail are picking up more exciting generative AI features. (link)
The AI hype train is distracting companies. (link)
NPR’s first podcast episode made by AI. (link)
Warren Buffet’s firm invests over $166B in 3 AI growth stocks. (link)
Tired of clickbait? Instagram founders’ new app re-writes clickbait article headlines. (link)
Baidu’s $145M AI fund signals China’s push for AI self-reliance. (link)
Tool of the Day ⚒️
AI tools we’ve used, loved, and highly recommend
Today’s tool is KTool.io.
If you’re an average person, you spend an average of 6hr58mins A DAY on LCD screens. Why does this matter? Because the “emissive” backlighting projected through LCD screens toward your eyes causes greater visual fatigue and sleep disruption than the “reflective” displays of Kindle.
With a 5-star rating from over 4,300 users, KTool lets you send news articles, Twitter threads, and newsletters directly to your Kindle, meaning you can slice through LCD screen time and do your reading on the far more eye-friendly Kindle.
We’ve found our knowledge retention and focus improve while using KTool.io as well, as we’re not receiving any notifications nor bumps from messages, emails, or other apps.
7 Questions
Interview with the founder of KTool
People making a go of it for themselves is sexy. Solopreneurs, developers, and small business owners; hat's off to all of them. Today, we’re stoked to share Daniel Nguyen’s story, the developer behind KTool.io and Boltai.app.
This is insight from someone who is walking the walk, so if you’re hoping to make something of your own, listen up; this could change your life.
To start, tell us a bit about your background, how you got into tech, and what you were doing before:
“I'm Daniel Nguyen. I'm an entrepreneur based in Ho Chi Minh City, Vietnam.
I love building things on the Internet. My first dollar I earned from the Internet was roughly 15 years ago when I was selling a plugin for an open-source forum software called Invision Power Board. Since then, I mostly do consulting & building side projects for fun.
Last year, I launched KTool—a productivity tool for Kindle owners. Recently, I launched BoltAI, an AI productivity tool for macOS users.”
You’ve developed BoltAI…in one sentence, describe this tool and what it does:
“BoltAI allows you to access ChatGPT in any Mac application. It integrates ChatGPT deeply into your workflows, so you don’t have to go back-and-forth with the ChatGPT web UI.”
Where does BoltAI stand now in terms of growth and your end goals? What made you start the project, what were the major hurdles, and what would you do differently if starting again from scratch?
“BoltAI is still fairly young, I launched the MVP in April. Currently, it has 280 beta users and 120 paid users.
When I started, I didn’t expect it to be commercially viable. It comes from my pet peeve of switching back and forth between ChatGPT web UI and native Mac apps like Xcode or Apple Notes.
So I decided to build the MVP in a weekend. The app was ugly but I managed to ship it anyway.
I launched on Twitter and got a few sales.
What would I do differently if starting again from scratch? Well, maybe to spend more time on the landing page, but it’s ok—I needed to ship fast.”
Since you've launched multiple products (boltai.app and ktool.io), what’s your process for coming up with these ideas, and how do you validate them?
“I have a list of all problems that come to my mind. Some can be a good opportunity, some are not, but writing them down helps me filter out the bad ideas.
Both KTool & BoltAI come from solving my own problems (the so-called “scratching your own itch” strategy).
I got lucky with both, as I spent little time on validation.
With KTool, I asked on Reddit and Twitter if people were looking for the solution I’m working on.
With BoltAI, I tweeted a few video demos, it went semi-viral, and many people signed up for early access.
Moving forward, I would build a super simple MVP within a week, share it on Twitter, Reddit, Indie Hackers & WIP.co, then listen to market feedback. If it gets enough traction, I will spend more time building it. Otherwise, I would move to another idea.
I shared my step-by-step process here.”
Has AI changed your life and how you work?
“Definitely. Although, I haven’t used many AI products yet. The product that is most impactful to me is ChatGPT (GPT-4 model). Fun fact: half of the BoltAI code in the MVP was from GPT-4.
ChatGPT helps me in many ways: brainstorm a new growth idea, draft a marketing plan, write SEO-optimized blog posts, analyze and debug application crashes…”
Was Twitter a big part of your success, and if yes, how do you suggest other people use and grow on Twitter?
“I think yes. When I started using Twitter more seriously last year, I immediately got overwhelming support from the founder & indie maker community. I’m probably not the best to talk about growing an audience on Twitter.
The “strategy” that works for me is to share my lessons on bootstrapping KTool & BoltAI. I occasionally create startup memes, and people seem to love these.”
Balancing life and indie-making can be a challenge. How do you maintain a work-life balance and how has this journey impacted your personal life?
“I’m still struggling with this. Earlier last year, we got a financial hit, KTool wasn’t growing, and we needed quick cash. I rushed to find consulting work while trying to fix KTool’s issues. I can’t say I had a work-life balance during that period.
But it’s getting much better now.
I think work-life balance should be a goal we keep in our mind to prevent burnout etc. But it’s not like, ok I worked 8 hours today, I should just stop.
I work more intensely when I’m inspired & motivated. And I would relax more when I feel exhausted and unmotivated.”
Thoughts?
We find insight like this, from people like Daniel, invaluable. Literally, life-changing stuff. Absolutely, Daniel clearly has a special knack for spotting holes in the app market, a developer skillset, and the motivation to follow through on fixing those problems. But hearing the step-by-step of how he comes up with these ideas, through to how he turns them into profitable apps, we absolutely love it. Let us know if you do too, and if enough do; we’ll continue these free interviews.
Spread The Word
Share and earn epic rewards
Randy’s gonna help us - will you?
Tell one person about our new newsletter design by sending them the link below ⏬.
If you do, we'll send you a totally free, extensive course on prompt engineering, where you will learn to:
Get to most out of GPT and Midjourney
Automate workflows
Free up your time
Do more cool sh*t instead of working