Reviewing Microsoft's New Bing AI Chatbot for Python Programming

Table of Contents

• Getting Access to the New Bing AI Chatbot
• Testing Bing AI Chatbot on Python Topics
• Assessing the Effectiveness of the Bing Chatbot
• Conclusion and Alternatives

Getting Access to the New Bing AI Chatbot

Microsoft has invested heavily in OpenAI and recently integrated a chat-like interface into Bing, similar to what you might be familiar with in ChatGPT. To get access, you have to join a waitlist. This is the recently updated version that has been heavily restricted by Microsoft due to several concerning responses when allowed to have long, context-rich conversations.

There are strict limits enforced to prevent questionable content. Conversations are wiped after about 5 queries, forcing you to start a new session. Bing can also end the chat itself if it detects the conversation veering into dangerous territory. In fact, if you simply ask Bing if its name is Sydney, it will immediately cut you off.

The Waitlist and Restrictions

Getting access to try out the new Bing chat functionality can be frustrating. You have to join a waitlist, and even then, access seems inconsistent. There are alsostrict limits enforced on the length of conversations and ability for Bing to retain context. After just a few questions, Bing will notify you that the conversation has reached its limit. This means you lose all context and have to start a brand new topic. It interrupts the flow of productive conversations.

Browser Compatibility Issues

Using Bing's new chat features also requires switching to the Edge Dev browser on certain platforms like Mac. The chat interface will prompt you to open Edge if trying to access it from Chrome or other browsers. You can still use regular Bing search, but lose access to the conversational aspects. There are also rendering problems even in Edge Dev at times. Some screens show up completely blank with corrupted text, preventing you from properly interacting with the chatbot. So the browser requirements and compatibility introduce additional frustrations.

Testing Bing AI Chatbot on Python Topics

Since Python is a popular programming language for developers and data scientists, I decided to test Bing's chat features in helping learn Python concepts and code examples.

Overall the responses were decent, but far from perfect. The chatbot can provide a reasonable starting point, but lacks the depth, context and reliability to fully replace human experts or dedicated learning platforms.

Python Course Suggestions

I started by asking Bing for a suggested 2-week outline to learn Python as a new developer. It provided a fairly thorough introduction, with week 1 covering basics like data structures and week 2 getting into classes and inheritance. A short quiz was also suggested after the first week. However, when I asked to see an example quiz, the formatting provided was difficult to follow. Code was intermixed with text rather haphazardly.

Python Code Examples

The code examples provided by Bing for simple use cases like a Flask API were reasonably well-constructed. I was able to test them out quickly and have them work as expected. Bing was also able to adjust the examples by adding attributes like 'type' when asked. However, for more complex requests like refactoring a long if statement, Bing struggled. When asking for an alternative solution, it simply provided back the original messy code instead of properly refactoring it.

Refactoring Python Code

I provided Bing with some deliberately bad code examples to see if it could properly refactor and optimize them. In some cases, it was able to provide cleaner solutions. But in other cases, it completely failed and just returned back similarly poor code. Bing also noted in a few examples that since it had never tested or run the code, it could not guarantee functionality. This differs from ChatGPT which will confidently assert code solutions as usable, even if they might not work in reality.

Assessing the Effectiveness of the Bing Chatbot

While Bing's new chat functionality shows potential, there are still many frustrating drawbacks holding it back as a daily driver replacement for search and programming assistance:

Firstly, integration with web search results seems lacking. Despite asking questions in a conversational format, unsatisfactory standard search results still appear mixed in with the chat responses. Secondly, having to constantly restart conversations and re-establish context severely limits productivity gains when trying to go deep on complex questions.

Search Result Issues

One major gripe is that traditional Bing web search results still populate when engaging with the chat, even if asking questions conversationally. For example, when asking for Python course outlines, an outdated Python 2 result appears as the top hit. Entering questions into the search box triggers inferior results compared to using proper search keywords in standard Bing. Integrating the natural language capabilities of the chatbot with tailored, high quality search remains an opportunity for improvement.

Conversation Restarts and Context Loss

Having conversations routinely terminated after just a few questions poses a major productivity barrier. Technical topics often require building up context across an entire dialogue. So constantly losing previous details wastes time re-explaining. It interrupts the natural flow of problem-solving, especially for more complex issues. Even adding one or two extra exchanges before resetting would likely allow more impactful usage of the chatbot.

Conclusion and Alternatives

Bing's new chat functionality is an intriguing first step in bringing natural language assistance into mainstream search. However, given the current limitations around question limits, browser compatibility, and search result relevance, I likely will not make Bing my go-to over other options... at least not yet.

Frustrations and Drawbacks

The need to constantly restart conversations, browser specific requirements, and lackluster default search integration all contribute to a somewhat frustrating initial experience chatting with Bing's AI capabilities. Having conversations routinely terminated after just a few questions poses a major productivity barrier for technical topics that require deeper contextual understanding.

Other Chatbot Options

For everyday search needs, traditional Google still provides a better integrated experience across devices and platforms. And for advanced conversational assistance, dedicated tools like ChatGPT have shown more reliable depth for technical subjects thus far. As Microsoft works to address the drawbacks over time in areas like question limits, browser support, and search integration, Bing may evolve into a more seamless portal for both information retrieval and natural language understanding.

FAQ

Q: Why does the Bing chatbot have restrictions?
A: Microsoft has restricted conversations to 5 queries before resetting to prevent potentially dangerous responses.

Q: What browser can I use the Bing chatbot on?
A: Currently only Microsoft Edge Dev allows access to the Bing chatbot.

Q: Does the Bing chatbot actually write working Python code?
A: Yes, the provided Python code examples generally work as expected.