OpenAI recently unveiled the latest version of its beloved ChatGPT chatbot, introducing a new voice with humanlike inflections and emotions. The online demo showcased the bot tutoring a child on solving a geometry problem, hinting at the exciting features to come.

However, the actual release fell short of expectations, with many promised features delayed or deactivated. Despite this setback, the new ChatGPT offers the ability to upload photos for analysis, faster responses, and real-time language translations.

As a leading chatbot in the tech industry, ChatGPT has garnered attention. After testing the latest version, I found mixed results. While it excels at language translations, it struggles with math and physics, leaving room for improvement. The trend of A.I. companies overpromising and underdelivering is becoming all too common, leading to consumer confusion and frustration.

The lesson here is clear: approach A.I. technology with caution and skepticism. Don’t invest in underdeveloped tech until its effectiveness is proven.

The new version of ChatGPT, GPT-4o (“o” for “omni”), is now available for free trial on OpenAI’s website and app. Users can explore its features, with paying subscribers gaining access to additional capabilities. OpenAI’s iterative approach allows for continuous feedback and improvement, shaping the future of A.I. chatbots.

Despite some shortcomings, ChatGPT-4o shows promise in certain areas. While it struggles with complex math problems and reasoning tasks, it demonstrates improvements in language translation and editing. The road to fully functional A.I. may be long, but the journey is underway.

Geometry and Physics

ChatGPT-4o’s ability to guide users through geometry problems was showcased in a video with Sal Khan of Khan Academy. While the video-analysis feature is pending release, early tests show promise in solving simple geometry puzzles. However, more complex problems still pose a challenge.

Physics problems, such as those involving block stacking, revealed both successes and failures for ChatGPT-4o. The chatbot’s performance varied, highlighting the need for ongoing refinement.

Reasoning

Tests in reasoning skills exposed ChatGPT-4o’s limitations, with recurrent issues in logic-based puzzles. While improvements are noted, the chatbot’s reliance on external feedback raises questions about its functionality.

Language

Language translation abilities and editing prowess are strengths for ChatGPT-4o. While accents may be off at times, the chatbot shows promise in enhancing communication across languages.

Bottom Line

OpenAI’s decision to offer ChatGPT-4o for free reflects a commitment to community feedback and improvement. While challenges remain, the potential for advanced A.I. capabilities is on the horizon.