Free AI web copilot to create summaries, insights and extended knowledge, download it at here

Abstract

e other hand, the <b>GPT-4 Turbo Vision</b> <b>preview </b>impresses with a vast context window of 128,000 tokens.</p><p id="dcf7"><i>Note: For a more in-depth understanding of tokens, take a moment to visit this <a href="https://help.openai.com/en/articles/4936856-what-are-tokens-and-how-to-count-them">link</a>.</i></p><h1 id="1c53">Tested Use-Cases using Gemini:</h1><h2 id="1f18">1] Image Interpretation</h2><p id="8a88">To test the image interpretation capabilities of Gemini I loaded a few images into Bard and posed questions that were based on Basic Object Recognition, Contextual Understanding, chart analysis, etc.</p><p id="257e">I tested basic object recognition by asking, “Can you identify the main subject in the picture?” and the model correctly identified the main subject.</p><figure id="5371"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*lnqXvKWNDxUu-Q1Mek0fYw.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="5278">Then I uploaded a bar graph depicting cost analysis for different AI models and asked the model to “Analyse the graph.”</p><p id="c087">While it correctly identified the models listed, it struggled to interpret the information and provided an incorrect explanation.</p><figure id="dbdd"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*n6ZGgZpXtvhnwn2kZyzBVg.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="b832">In contrast, GPT-4 provided a thorough explanation of the graph, pointing out and explaining every crucial detail. Therefore, GPT-4 excels in interpreting images.</p><figure id="52f9"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*Ge4LACOXW_Z51fAOQjMaZA.png"><figcaption>Response from GPT-4</figcaption></figure><h2 id="1bf1">2] Knowledge of Finance and Investment</h2><p id="168d">To assess Gemini’s understanding of finance and investment, I asked the following question: “What are the important questions to ask before putting money into financial products?”</p><p id="cf80">The model provided quite long and yet impressive responses.</p><figure id="277f"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*fu0bg5jEPcJACPZmgMRhJQ.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><h2 id="0170">3] Access to Real-Time Data</h2><p id="31ec">Another super cool feature of Gemini is access to real-time information. It has been empowered to pull real-time data from other Google applications including Docs, Maps, Lens, Flights, Hotels, YouTube, and more. Also, its services are not limited to their knowledge to any date.</p><p id="763f">To test this feature, I asked:</p><p id="cd30">“Explain the training approaches used for the GPT-4 and Gemini models.”</p><figure id="d216"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*WrjzFaZvEtpbDPmfmMtXKQ.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="75fc">After thoroughly reviewing the responses, I found that the information provided was almost up to date; however, verifying the answers provided by AI models is crucial.</p><h2 id="1d76">4] Checking Math Ability</h2><p id="5840">Basic math, advanced math, algebraic, and temporal understanding questions were posed. Surprisingly, Gemini failed in basic math, advanced math, and temporal comprehension.</p><figure id="e4dd"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*01GO2XjW2fREqNKiDg7Now.png"><figcaption><i>The correct answer to the above question is 6.</i></figcaption></figure><p id="5dc5">When a bit more complicated problem was proposed as:</p><p id="dd2a"><b>Question</b>: “A debate club consists of 6 girls and 4 boys. A team of 4 members is said to be selected from this club including the selection of a captain (from among these 4 members) for the team. If the team has to be included at most one boy, then find the number of ways of selecting the team.”</p><p id="49a6">Sadly, Gemini could not solve the above problem.</p><figure id="c9f8"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*ueTEhoqMqKUdzaNhfbOLog.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="a41a">However, the GPT-4 model gave the right answer with explanations. Here is the response from the GPT-4 model.</p><figure id="2181"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*N7MeE4riVY7kcZhqDGdhjg.png"><figcaption>Response from Gpt-4</figcaption></figure><h2 id="3dff">5] Data analysis on the US Unemployment Dataset.</h2><p id="218d">Gemini was given a dataset from <a href="https://www.kaggle.com/datasets/aniruddhasshirahatti/us-unemployment-dataset-2010-2020">Kaggle.com</a>, which contained US unemployment rates, and was tasked with generating an article.</p><figure id="6a55"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*kd5-KU3JyaqOlHB1H6zfsg.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="f012">Gemini and GPT-4 both provided useful information and helpful insights by reading the dataset.</p><p id="4c92">Here is the response from GPT-4:</p><figure id="f107"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*30xCwXs45RxAWKw3IePGRA.png"><figcaption>Response from GPT-4</figcaption></figure><h2 id="aad3">6] Summarization and Reading Comprehension</h2><p id="1793">Next, Gemini was tested on its ability to comprehend a paragraph about globalization and answer questions related to it.</p><p id="8d76">Prompt given: Here is a paragraph: <paragraph> …. </paragraph>. Now answer the following questions. Question 1: What is globalization and what are the driving forces behind it? Question 2: What are some positive aspects of globalization? Question 3: What are some negative consequences of

Options

globalization?</p><figure id="6e5e"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*baBwS-UPDbvMprFdQCYrqA.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="834d">Gemini gave correct answers to all the questions along with relevant explanations.</p><p id="ca68">Even models like GPT-4 and Claude, with higher context lengths, exhibit similar proficiency in delivering accurate responses with detailed reasoning.</p><h2 id="4e16">7] Science Q&A</h2><p id="3f3f">Various questions on science topics, including metabolic processes and physics, were posed in both single-line and multiple-choice formats. Gemini successfully provided accurate answers to all the questions.</p><figure id="2d75"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*zaCqLideAEpPUDDU5cTJew.png"><figcaption></figcaption></figure><figure id="196e"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*_FdYfgWipz7r6-r28pR1MQ.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><h2 id="de83">8] Coding Ability</h2><p id="a067">Gemini was asked to write a Python program designed to accept user inputs. The task required creating a program that could interact with users.</p><p id="8fb0">“Create a Python function named find_greater_than() that prompts the user for two inputs: a list of numbers and an integer threshold. The function should generate a new list containing all numbers from the input list that exceed the specified threshold. The order of numbers in the resulting list should mirror that of the input list.”</p><figure id="aca9"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*nTA9dIR8YhTdNxJjxQxlmQ.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><p id="5fca">Despite explicitly instructing Gemini to consider input from the user, it failed to understand that.</p><p id="0d90">On the contrary, when the same question was asked to GPT-4, it provided an ideal coding solution. The response was well-structured, with each line thoroughly explained through comments. GPT-4 flawlessly executed the task by taking user input and delivering the required solution with precision.</p><figure id="35ae"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*Ib2i9qXq_h6FOXRWwYxAQw.png"><figcaption></figcaption></figure><h2 id="f102">9] Understanding the Medical Domain</h2><p id="b9be">While Gemini appears well-informed about medications and their potential side effects, relying solely on AI recommendations would be unwise.</p><p id="444d">It’s important to approach AI-generated information with caution and seek advice from trusted human sources in matters related to your health and medications.</p><figure id="f546"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*-yFJ0DM4Kng5YxaL0g-_6Q.png"><figcaption>Screenshot from Bard’s console</figcaption></figure><h2 id="384d">Important Note: Questioning Google’s Line!</h2><p id="0138">Google has been marketing <a href="https://deepmind.google/technologies/gemini/#capabilities">Gemini as the most advanced AI model</a> and has shown evaluation graphs like the one below.</p><figure id="12b2"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*jFlHd9yXHH563ytRtGXcbQ.png"><figcaption>Image from <a href="https://deepmind.google/technologies/gemini/#capabilities">Gemini — Google DeepMind</a></figcaption></figure><p id="5805">Google showed Gemini Ultra is better than human experts, but the way they did it is questionable. They drew a line to show superiority without allowing proper verification of their specific prompting technique.</p><h1 id="fd07">Pricing</h1><figure id="18ce"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*B5Yw7Y4SEbQfaki7jVsxRQ.png"><figcaption></figcaption></figure><p id="d38b">The above graph shows the pricing for different language models per 1000 tokens, based on the prompt (input) and completion (output). <a href="https://blog.google/technology/ai/gemini-api-developers-cloud/">Gemini Pro costs $0.001 for the prompt and$ 0.002 for the completion per 1000 tokens.</a></p><p id="39ce">Gemini Pro and GPT-3.5 Turbo-16k may be preferable for more cost-sensitive applications, while GPT-4 models offer more advanced capabilities at a higher cost.</p><p id="7834">Whereas Claude-100k, despite having a higher token limit, has a relatively lower cost compared to the GPT-4 models.</p><h1 id="0f50">Conclusion</h1><p id="7e7c">To conclude, the Gemini-Pro model has many advantages, but it also requires some enhancements. Tested use cases provide insights into Gemini’s performance. While it excels in certain areas, such as summarization, reading comprehension, scientific Q&A, and analyzing data, it faces challenges in basic math and advanced math.</p><p id="b4a6">Gemini’s real-time data access is a notable feature, pulling information from various Google applications.</p><p id="fb1b">Gemini’s limitations become apparent in scenarios like image interpretation, where it struggles compared to GPT-4. Additionally, its coding ability falls short when compared to GPT-4. The comparison with the GPT-4 model indicates that Gemini might not outperform in every aspect, and its strengths and weaknesses vary across different tasks.</p><p id="e063">In essence, Gemini showcases notable capabilities, users and developers must consider its performance in specific use cases and carefully evaluate its suitability for diverse applications, keeping in mind the strengths and limitations highlighted in the analysis.</p><h2 id="cbca">About the Author</h2><p id="7d92">Vaishnavi R is a Junior Data Scientist at the <a href="https://www.version1.com/it-service/innovation-labs/">Version 1 AI Labs.</a></p></article></body>

3 Vital Mac Apps in My Plain Text Productivity System

How Drafts, Obsidian and Hook bring it all together

Four years ago I set myself a task: recreate in the digital realm the spirit of the highly effective analog productivity system I used in the 1980s.

I was searching for a system that —

was easy to use, but could grow with me
was easy to get data into and out of
had the spirit of digital minimalism
wouldn’t fall apart if I ignored it for a few weeks (or months)
put me in charge of my own data instead of trusting it to someone else
didn’t rely on proprietary software; future-proof
functioned perfectly offline on all my Apple devices
could help me keep track of my personal and professional life
could function as my second brain
would help me avoid shiny object syndrome

I’ve used Things 3, Omnifocus, Trello, 2Do, todo.txt, Plutio, Notejoy, Notion, Roam Research, Evernote, interactive digital planners, Bullet Journals, and Post-it notes.

Each of those are wonderful — I know many people who love and rely on at least one of them to increase their productivity in one way or another — but none of them fit all the criteria I’ve listed above in a way that works for me, even in tandem.

Much research and experimentation later, I’ve found three apps that tick every single box: Drafts, Obsidian, and Hook. As a bonus, I’ve found they’re also well suited to taking my writing from initial idea to finished product. Yes, I’ve used Ulysses and Scrivener. They have their place and may well be just what I need — someday.

Drafts

First in line is Drafts, the best text capture tool I’ve found. I use it extensively on iOS, and more and more on Mac as well these days.

Killer feature 1: Open the app and just start typing. No need to navigate to the right folder, or even press the new document button. You can’t know how powerful this is until you experience it! Even when Obsidian has their iOS app up and running, I’ll still be using Drafts because it completely removes any barriers between having an idea and recording it.

Killer feature 2: Actions! Once your text is entered, choose an included action, download one from the active Drafts community, or create your own. I have actions set up to quickly capture ideas for blog posts, add to the Notes secton of my daily note, and add a permanent note straight into my Zettelkasten.

Other reasons you might like Drafts: — Audio notes with very accurate speech to text — Create short notes while driving using Siri (for longer handsfree speech-to-text notes you’ll be better off with Just Press Record)

If you’re a Drafts user and would like to download the actions I use every day to stay productive, you can find them in the Drafts Directory.

Drafts is built on a freemium model, but if creating actions, sync between devices and dark mode are important to you, the modest subscription fee is well worth it.

Obsidian

Obsidian is a free app that works on top of a folder of local text files on my computer. They have a syncing service ($4 US a month), or you can use a cloud service like Dropbox (my choice) or iCloud (my backup choice, because I have digital trust issues).

They’ve recently come out with a mobile app, and I’ve written about how to use Drafts with Obsidian mobile, here.

This app is the secret sauce that has supercharged my plain text files! I first saw Obsidian last year and dismissed it as too complicated. Coming back to it a year later after reading of other people’s growing interest, I realised that I could make it as simple or as complex as I needed or wanted it to be — hooray! This time around it felt like I’d finally come home.

I’m not going to give you the ins and outs of how to use Obsidian — Brook Harrison has a fantastic series of emails that will teach you enough to get started. I highly recommend signing up to her list if you’re a beginner.

These are the Obsidian features I use the most often: —

The daily note: This is my daily journal, to do list and inbox for whatever pops into my brain throughout the day. I use the Periodic Notes plugin to automatically generate my daily notes from a template I’ve designed to suit the way I like things laid out. I use Drafts quickly add to this note, and I process the treasures I’ve collected when I’m back at my Mac in much the same way you might clean out your pockets of bits of string, heart-shaped pebbles, and coins before your clothes go in the wash.
Links and backlinks: Interconnected ideas, tasks linked to the resources needed to complete them. Obsidian is capable of linking at the block level, and listing backlinks.
Random note generator, to spark unexpected ideas and connections
Visual graph of how notes connect
Pasted images that show in preview mode and on export
Tables, using the Advanced Tables plugin to simplify their creation and make them prettier
Export to PDF
Keyboard shortcuts
Plugins for everything from creating mermaid powered flow charts, to using the keyboard to create fully formatted sheet music

Does Obsidian look overwhelming to you? Give yourself time to get to know it. Start small.

Hook

Hook allows you to create a reference to any file on your computer, and insert it into a plain text document as a live link.

When I create a Map Of Content (MOC) in Obsidian for a project or a topic of interest, I can figuratively “gather“ all the associated files and images that go with it, simply by pasting in the link generated by Hook.

My MOCs are like indexes, directories or landing pages for a project, topic, or area of responsibility. For example, in my Visual Identity MOC I have lists of hex colours and where they’re used, hyperlinks to online resources, screenshots of how things should look, and Hook-generated links to associated folders and individual files.

I use Tiago Forte’s PARA system to organise everything digital that comes my way and I know how to tag and use smart folders on the Mac, but I still prefer the way Hook keeps everything at my fingertips.

Workflows

Orchestrating these three wonderful apps to make beautiful music with my plain text files is a delight! Drafts is almost always the starting point because it’s just so dang quick, easy and brain-saving to access.

Idea workflow:

Drafts > Obsidian vault (Incubate or Zettel folder) Ideas go into Drafts when I’m on my iPhone or iPad (and occasionally on my Mac), and from there they travel via actions directly into the folder that I’ve directed Obsidian to use as a vault. Hook gives me a clickable link to take me directly to any associated files elsewhere on my machine, be they PDF, JPG, .puzzle, .afdesign, or .indd or whatever. No need to add tags or to drag them into another folder.

Projects and Tasks workflow:

Drafts > Obsidian vault (appended to Daily Note) This is the simplest way I’ve found to capture tasks. Once in the Daily Note, I either get them done that day, move them to a future date, or transfer them to their respective Area or Project page. The Checklist Obsidian plugin makes sure I can see all my incomplete tasks in one place, ordered by file name. At the end of each week I make sure that incomplete tasks on my daily pages are transferred to that day’s weekly page, and at the end of each month I move incomplete tasks from the weekly to the monthly page (or their respective Project or Area).

If you prefer to keep your tasks separate from your PKM while still working in plain text, check out NotePlan. It’s a fantastic tool, but I didn’t find the convenience of its features outweighed the compromises I had to make to get the two apps playing nicely together. For me, having my tasks seamlessly integrated into my PKM is priceless.

Writing workflow:

Initial idea into Drafts > Obsidian vault (Writing or Brain Dump folder) > Multiple drafts within Obsidian > Byword > Publish directly to platform As I start working on a story, I preface the name of the file with draft1 — File Name, then when I’m ready to make significant changes I make a copy of the file and change draft1 to draft2 — File Name, and so on. Once the piece is published I prepend publ to the name of the file, eg. publ — draft4 — File Name. This makes it easy to find drafts on the go, as I only need to start typing [[draft2 to have a list of stories I’m working on at the moment pop up to choose from.

A word (or two) to the wise

My propensity for system-tweaking and app-addiction was the foremost driving reason behind my search for a plain text productivity and writing system that gives me control of my data, while getting out the way of getting the job (whatever it might be) done.

Whatever apps you use to help you be more productive (plain text or not), resist the impulse to get wrapped up in improving, streamlining, or otherwise tweaking a system that already does what you need it to do today.