Google announced that there would be a limited release of Bard and they have followed through with this information allowing individuals who were on the waitlist to finally test out their new artificial intelligence product Bard. Bard is a product Google has long had in development, but has fast tracked due to the commercial success of competitor OpenAI’s Chat-GPT product. Both Chat-GPT and Bard utilize artificial intelligence to provide information in a conversational way. At a base level both of these products act as search engines, but provide information in a conversational way. These products are also able to rewrite information which it has gathered from around the web to provide unique and valuable responses to what users type in.
First Look at Google’s Bard Interface
The image above is taken directly from Google’s Bard interface. The interface itself, like all Google products, is a minimalistic clean looking format. Above you are able to see the chat-like interface which allows users to interact with Bard. In this case, we gave Bard a simple hello and asked about the local weather in Philadelphia. As you can see in the response above, Google utilizes the same information found within search results to then generate an answer which sounds conversational.
The above example is an easy test for Google’s Bard. One of the biggest advantages to artificial intelligence products like Bard is the ability to “think” and provide unique or complex answers. Let’s take a look at another example.
For our next example, we wanted to ask Google a question that was a little bit more complex. Of course, Bard was easily able to answer the question “What is quality score in Google Ads?”. The next test for Bard was to see if it was able to parse the information about Google Ads quality score to just provide me with things that I could do to improve my quality score.
Google’s Bard passed the test with flying colors. It was able to provide information from Google’s support documents for Google Ads, while just including the information relevant to improving my quality score.
Limitations of Artificial Intelligence Products
There are of course limitations of these exciting new artificial intelligence products which users should take caution against. As with any machine learning mechanism, the more information which goes into the system, the more accurate it will continue to get. With both Chat-GPT and Bard being relatively new technologies, there is still a lot of learning for these machines to do in order to meet their full potential. Not only is this true for the complexity for answers provided, but also the accuracy. Let’s view the example below.
We asked Bard if Philadelphia Phillies 1st baseman Rhys Hoskins was good. The reason we utilized this for an example is that the answer is subjective. We can not prove entirely whether the opinion being generated is true.
In the above example, Bard pulls stats for Rhys Hoskins’ career and determines that based on these statistics it reasons that Rhys is a good 1st baseman. While this portion of the answer may be debatable for baseball fans, the second paragraph is an inaccuracy which is undeniable.
The second paragraph indicated Rhys Hoskins tore his hamstring in 2021 and had a knee injury in 2022. Bard then reasons that a concern regarding Hoskins may be his injury history. Unfortunately for Bard, Rhys Hoskins never tore his hamstring during the 2021 season. During the 2021 season Rhys tore an abdominal muscle and his hamstring was never a concern. Bard was accurate when describing that Rhys was “hampered by a knee injury” in 2022. This an impressive determination by Bard since Rhys never missed considerable time during the season for the injury, but did have offseason surgery to repair a torn meniscus.
Let’s take a look at Bard’s ability to think and drill down on topics. One of the features which Google promotes is Bard’s ability to assist with blogs. In the below example, I ask Bard for some ideas for our digital marketing blog.
As you can see in the above example, the first answer from Bard was not very specific. Bard provided me with some topics for blogs, but these blog topics are more of categories than a topic which a specific blog can be written about.
I then ask Bard to be more specific in the topics which it provides to me, the results it provides are great ideas for blogs for a digital marketing company. These ideas are relevant and useful to end users. Being able to develop ideas and topics like this instantly is one of the great benefits of Bard and related artificial intelligence products. Generating information in this way can greatly increase efficiency for businesses and reduce costs.
Why Artificial Intelligence Inaccuracies Are Dangerous
An answer to whether an individual baseball player is good is not dangerous on its own, however; inaccuracies within artificial intelligence can be dangerous in certain circumstances. Due to the authoritative manner in which answers are provided, users are more likely to believe the answer being provided without questioning it. In this case disinformation can quickly spread as users go to what they expect to be an accurate source consistently.
While these artificial intelligence products will continue to learn and get more accurate, it will continue to be a battle to ensure inaccuracies are quickly addressed. Artificial intelligence products all utilize inputs which could be manipulated to increase or decrease visibility for specific information. While this sounds nefarious, this is much in the same way digital marketing companies manipulate search engine inputs to increase visibility for their clients by completing search engine optimization services. Information from artificial intelligence products like Chat-GPT and Bard should face the same scrutiny as information presented in search results during the advent of Google. As these machines continue to learn, accuracy will improve, just as the quality of Google’s search results have.
Results Guaranteed will continue to monitor Bard and other artificial intelligence tools. These tools will have great overlaps with digital marketing from the standpoint of streamlining their own internal process, but also changing how they attempt to gain visibility for their clients. At this point, Bard and Chat-GPT are in their infancies and will see massive changes in the months and years to come.