I have talked about OpenAI, Dall-E and ChatGPT before in bits, how Bing has integrated with AI, and where AI is heading now. But there is one AI tool that I have not been able to review due to geolocation restrictions and area of testing – until I had a workaround with VPN. That is Bard by Google.
Not long after Microsoft announced its collaboration with OpenAI on February 7th, 2023, Google announced their own AI called Bard that same week. Bard is a language model developed by Google. The model is based on the GPT-3 architecture and has been trained on a massive amount of data, enabling it to generate human-like language responses.
 |
| Source: Google |
Over the years, Google is known to have a variety of AI tools and Machine Learning Algorithms based on collective data and random datasets. Bard is based on a lightweight and optimized version of LaMDA (short for "Language Models for Dialogue Applications"), and, like most LLMs today (Large Language Model), was pre-trained on a variety of data from publicly available sources.
One of the key features of Bard is its ability to perform tasks such as summarization, translation, and question-answering with a high level of accuracy. It can also generate coherent and engaging text on a wide range of topics, making it a useful tool for content creation.
Another noteworthy aspect of Bard is its focus on responsible AI. Google has made efforts to ensure that the model is free from bias and can generate content that is respectful and appropriate. The company has also implemented measures to prevent the model from generating harmful or misleading information. These are yet to be tested in the experimental phase.
Hands-on Interaction.
Since the test preview is available to US and Canada only, I had to test out Bard on April 15th via VPN (Psiphon Pro app on mobile and Opera's inbuilt VPN on PC). It did not take long to set up my account because all I had to do was join the waitlist – which took like three to five minutes from "You're on the waitlist" to get an email saying, "It's your turn to try Bard." Nice.
 |
| Left: Email waitlist confirmation. Right: Email invitation to try Bard. |
I have already logged in to my Google account and all I can do to access Bard is to search Bard or by visiting bard.google.com. Now I do not need to turn on VPN every time I am visiting Bard, so that is a plus (since Google thinks I am in the States).
What you need to know is that Bard has slightly improved in the past months, so these screenshots will show how each of the results is displayed. These results are based on random results matching the topic or question.
 |
| Starting off with a brief intro. Bard gets to describe itself. |
.png) |
| Got curious to see if Bard would deliver this: Creating Images. |
.png) |
| Sadly, Bard still can't create AI images. Just gives descriptions of images in words. |
 |
| Bard can provide drafts of information you ask based on the topic or keyword. The details can be well summarized. |
.png) |
| At the end, you can see the Google It option, used just to confirm whether the information provided is accurate or not. |
Before April, testers were unable to get programming results by asking questions to generate code. But now, you can do that. Bard can now write code in any language like Python, JavaScript, Java and C++.
.png)
.png) |
| Bard can now write code in any language. I chose C++ as a test. |
Either way, Bard works similarly to Bing AI and ChatGPT. Depending on how one uses it, Bard will deliver the result to the user instantly.
Concerns with the AI model
Weeks after Bard was announced, other reviewers and beta testers have been met with similar concerns of wrong information, hate speech, toxic content and shallow or half-baked information. Of course, this made Google aware of the issues and later they admitted that their data and information is not 100% accurate and appropriate. So, the footnote on the Bard page and first-time greeting message will always display: Bard may display inaccurate or offensive information that doesn’t represent Google’s views.
.png) |
| Look at the centre message and footnote. |
Closing Remarks
Google still has a long way to go to catch up with ChatGPT because its product announcement seemed to be rushed to compete with Microsoft in my opinion. The developers still need to keep working on it since it is still in the experimental phase.
Overall, Bard is a promising addition to the field of language models and has the potential to be a valuable tool for a wide range of applications. Its ability to generate high-quality text and its focus on responsible AI make it a model worth keeping an eye on in the coming years.
