Testing Google’s Gemini-Exp-1206 model in data analysis and visualization

One of Google’s latest experimental model, Gemini-Exp-1206, It shows the potential to mitigate one of the harshest aspects of all analyst Function: Get their data and visualizations perfectly in sync and deliver a compelling narrative, Without having to work all night.

Investment analysts, junior bankers, and members of advisory teams who aspire to partner positions take their roles and they know it. Long hoursAnd weekends and occasional all-nighters can give them an inside advantage in the promotion.

What consumes a lot of their time is performing advanced data analysis while also creating visualizations that enhance Compelling story. What makes this even more difficult is that each banking, fintech, and consulting firm, such as JP Morgan, McKinsey, and PricewaterhouseCoopers, has unique formats and agreements for analyzing and visualizing data.

VentureBeat interviewed members of internal project teams to which these companies’ employers hired and assigned the project. Producing visuals that condense and integrate the vast amount of data is an ongoing challenge, said employees who work in consultant-led teams. One said it was common for consulting teams to work overnight and do at least three to four iterations of presentation visualizations before settling on one and preparing it for board-level updates.

A compelling use case for testing Google’s latest model

Process analysts rely on creating presentations that support a story with powerful visualizations and graphics containing many manual steps and iterations that have proven a compelling use case for testing Google’s latest model.

When the model was launched earlier in December, Google’s Patrick Kane books“Whether you’re tackling complex programming challenges, solving mathematical problems for school or personal projects, or providing detailed, multi-step instructions to formulate a customized business plan, Gemini-Exp-1206 will help you navigate complex tasks more easily.” Google noted the model’s improved performance on more complex tasks, including mathematical reasoning, programming, and following a sequence of instructions.

VentureBeat took the Google Exp-1206 model for a comprehensive test drive this week. We have created and tested over 50 Python scripts in an effort to automate and integrate intuitive, easy-to-understand analysis and visualizations that can simplify the complex data being analyzed. Given how hyperscalers dominate today’s news cycles, our specific goal was to create an analysis of a specific technology market while also creating supporting tables and advanced graphics.

Through over 50 different iterations of verified Python scripts, our findings included:

• The more complex the Python code request, the more the model will “think” and try to predict the desired outcome. The Exp-1206 attempts to anticipate what is required from a given complex router and will change what it produces with even the slightest change in the router. We saw this in how the model alternated between table type formats placed directly on top of the Hyperscale Market Analysis spider chart we created for testing.

• Forcing the model to attempt to analyze and visualize complex data and produce an Excel file results in a multi-tab spreadsheet being presented. Without ever being asked to have an Excel spreadsheet with multiple tabs, Exp-1206 has created one. The basic tabular analysis required was in one tab, the visuals in another, and an additional table in the third.

• Telling the model to iterate through the data and recommend the 10 visualizations that it decides fit the data best produces interesting and useful results. In order to reduce the time drain of having to create three or four iterations of slide decks before reviewing the board, we forced the model to produce multiple conceptual iterations of the images. They can be easily cleaned up and integrated into your presentation, saving many hours of manual labor creating charts on your slides.

Exp-1206 pushed towards complex, multi-layered missions

VentureBeat’s goal was to see how far the model could be pushed in terms of complexity and multi-layered tasks. Its performance in creating, running, editing, and tuning 50 different Python scripts demonstrated how quickly the model attempts to capture nuances in code and interactivity on the fly. The model flexes and adapts based on immediate history.

Result of running Python code generated with Exp-1206 in Google Colab Show that fine detail extends to the shading and transparency of layers in an eight-point spider graph designed to show how six ultra-compact competitors compare. The eight attributes we asked Exp-1206 to identify across all supermeters and spider graph installation remained consistent, while the graphical representations varied.

Battle of Hyperscalers

We selected the following hyperscalers to compare in our testing: Alibaba Cloud, Amazon Web Services (AWS), Digital Realty, Equinix, Google Cloud Platform (GCP), Huawei, IBM Cloud, Meta Platforms (Facebook), Microsoft Azure, and NTT Global Data. Hubs, Oracle Cloud, and Tencent Cloud.

Next, we wrote an 11-step prompt containing over 450 words. The goal was to see how well the Exp-1206 could handle sequential logic and not lose its place in a complex, multi-step process. (You can read the prompt in the appendix at the end of this article.)

We then sent the claim in Google Artificial Intelligence Studiochoosing the Gemini demo model 1206 as shown in the figure below.

Next, we copied the code to Google Colab, saved it in a Jupyter notebook (Hyperscaler Comparison – Gemini Experimental 1206.ipynb), and then ran the Python script. The script ran flawlessly and created three files (indicated by red arrows in the upper left).

Comparative analysis and graph of Hyperscaler – in less than a minute

The first series of instructions in the prompt asked Exp-1206 to create a Python script that would compare 12 different hypermeters by their product name, unique features, differentiators, and data center locations. Here’s how the required Excel file appears in the script. The spreadsheet format took less than a minute to shrink to fit the columns.

The next series of commands asked for a table of the top six hyperscalers compared to the top of the page and the spider graph below. Exp-1206 alone chose to represent the data in HTML, creating the page below.

The final spot order sequence focuses on creating a spider graph to compare the top six hyperscalers. We tasked Exp-1206 with selecting the eight benchmarks to compare and complete the plot. This command string was translated into Python, and the model generated the file and presented it in a Google Colab session.

A model specifically designed to save analysts’ time

VentureBeat has learned that in their daily work, analysts continue to create, share, and fine-tune claims libraries for specific AI models with the goal of simplifying reporting, analysis, and visualization across their teams.

Teams dedicated to large-scale consulting projects need to consider how models like the Gemini-Exp-1206 can dramatically improve productivity and alleviate the need for 60+ hour work weeks and occasional night hours. A series of automated prompts can do the exploratory work of looking at relationships in data, enabling analysts to produce pictures with greater certainty without having to spend a significant amount of time getting there.

Trailing:

Google Gemini Demo 1206 Quick Test

Write a Python script to analyze the following hyperscalers that have announced global infrastructure and data center presence for their platforms and create a comparison table between them that captures the important differences in each approach to global infrastructure and data center presence.

Let the first column of the table be the company name, the second column be the names of each of the company’s hyperscalers that have a global infrastructure and data center presence, the third column be what makes their hyperscalers unique and dive deep into their most distinctive features, and the fourth column be locations Data centers for every hyperscale at the city, state and country levels. Include all 12 Hyperscalers in the Excel file. Don’t throw away the web. Create an Excel file of the result and format the text in the Excel file so that it is free of any parentheses ({}), quotation marks (‘), double asterisks (**) and any HTML code to improve readability. Name the Excel file, Gemini_Experimental_1206_test.xlsx.

Next, create a table three columns wide and seven columns deep. The first column is titled Hyperscaler, the second is titled Unique Features and Differentiators, and the third is Infrastructure and Data Center Locations. Column headings are bolded and centered. Hyperscalers also bold addresses. Double-check to make sure that the text inside each cell in this table wraps around itself and does not intersect with the next cell. Adjust the height of each row to ensure that all text fits in the intended cell. This table compares Amazon Web Services (AWS), Google Cloud Platform (GCP), IBM Cloud, Meta Platforms (Facebook), Microsoft Azure, and Oracle Cloud. Center the table at the top of the output page.

Next, take Amazon Web Services (AWS), Google Cloud Platform (GCP), IBM Cloud, Meta Platforms (Facebook), Microsoft Azure, and Oracle Cloud and identify the eight most distinctive aspects of the group. Use these eight distinct aspects to create a spider graph comparing these six superior metrics. Create one large spider graph that clearly shows the differences in these six supermeters, using different colors to improve readability and the ability to see the outlines or footprints of the different supermeters. Check the title of the analysis, What’s the Best Hyperscalers, December 2024. Make sure the legend is fully visible and not on top of the graphic.

Add a spider graphic at the bottom of the page. Center the spider graphic below the table on the output page.

These are the super scalers that should be included in a Python script: Alibaba Cloud, Amazon Web Services (AWS), Digital Realty, Equinix, Google Cloud Platform (GCP), Huawei, IBM Cloud, Meta Platforms (Facebook), Microsoft Azure, NTT Global Data. Hubs, Oracle Cloud, Tencent Cloud.