AI: Chat GPT vs. Google Bard battle.
By Juan Pablo Romano, Engineering Manager at Adviters.
Currently, Artificial Intelligence (AI) has become a topic of great importance. AI is a field of study that focuses on developing systems and programs capable of performing tasks that would typically require human intelligence. Its impact spans various areas of our lives, from virtual assistants on our devices to revolutionary medical advancements. In this article, we will explore what AI is and its crucial role in our current world.
Advancements in the field of AI have been impressive in recent years. Researchers and scientists have succeeded in developing increasingly sophisticated algorithms and models that enable machines to learn and make autonomous decisions. Machine learning has become a fundamental pillar of AI, where algorithms are trained with large amounts of data to recognize patterns and make informed decisions.
Furthermore, Natural Language Processing (NLP) has made significant progress, allowing machines to understand and generate text more accurately. These advancements are driving the adoption of AI in a wide range of applications, from healthcare to the manufacturing industry.
Does NLP sound familiar? And to think that you used ChatGPT all day today.
So, what is artificial intelligence, then?
Alan Turing, the creator of the first computer and the father of modern computing, is also recognized as the creator of artificial intelligence (he created a very cool test called the «Turing Test») and defines AI as a discipline that seeks to develop systems and programs capable of performing tasks that would normally require human intelligence. Therefore, these systems are designed to learn, reason, perceive, and make decisions autonomously. AI is based on algorithms and models that mimic the way humans process information and make decisions, allowing machines to emulate certain aspects of human intelligence.
Within the field of artificial intelligence, there are various approaches used to achieve different types of intelligence in machines. One of the most prominent approaches is machine learning, which is based on the ability of machines to learn from data without being explicitly programmed.
In summary, AI is defined as the field of study that seeks to create systems and programs capable of imitating human intelligence.
Now that we know what we call Artificial Intelligence, we can ask ourselves: what challenges and considerations should we have regarding this technology that we created to imitate human intelligence?
To answer this question, we have to consider four very important points: data gap and available data quality, algorithmic bias, and ethical challenges in the use of AI.
Therefore, in the development of Artificial Intelligence, it is crucial to have high-quality and representative datasets. However, there is a significant data gap in terms of access and quality. This can lead to biases and limit the ability of AI systems to make accurate and fair predictions. It is important to address this gap and promote the availability of diverse and representative data to avoid discriminatory biases.
On the other hand, we have what is called «Algorithmic Bias» (Bard and ChatGPT are always affected by this bias), which is inherent in the training data. This can result in discriminatory or biased decisions. This means that if we train an AI to be good using data that is corrupted in some way, the content it generates will not be good. It is essential that human intervention be present during the training of AI to correct this bias (besides, no one wants to be the creator of the real Skynet, right?).
All of this presents us with certain complex ethical challenges in the creation and use of AI, such as the impact on human employment and transparency in the use of algorithms to protect the data and information with which we train AI.
We’re tired of theory, aren’t we? Well, let’s go with…
A list of platforms that use Artificial Intelligence and can assist you in work or study.
- OpenAI ChatGPT
Natural Language Processing
- Google Bard
Natural Language Processing
- OpenAI Dall-E2
Image Generation
- Midjourney
Image Generation (with better quality compared to Dall-E)
- BlackBox
AI dedicated to generating or writing source code based on textual queries in English
- Caiyon
Image Generation
- Github
Copilot AI that installs as a plugin in the IDE and «assists» or suggests code blocks based on reading your code or textual queries in multiple languages.
Lastly, before the conclusions, let’s ask ChatGPT to compare itself to Bard and Bard to ChatGPT.
ChatGPT
Prompt: Can you make a comparison between yourself and Google Bard?
ChatGPT: Of course, I can make a comparison between myself (ChatGPT) and Google BERT.
Underlying Model:
Me (ChatGPT): I am based on the GPT (Generative Pretrained Transformer) language model developed by OpenAI. GPT is a transformer architecture that focuses on generating coherent and contextually relevant text.
Google BERT: BERT (Bidirectional Encoder Representations from Transformers) is a language model developed by Google. It is based on the transformer architecture and is designed to understand the context and relationship between words in a sentence.
Capabilities:
Me (ChatGPT): I am designed to generate responses and maintain coherent conversations with users. My main objective is to provide relevant information and assist in various natural language tasks.
Google BERT: BERT stands out for its contextual understanding ability and its skill in performing specific natural language processing tasks such as text classification, information extraction, and text generation.
Training and Data:
Me (ChatGPT): I was trained using an extensive dataset that includes a wide variety of texts from different domains. I learned in a generative manner, trying to predict the next word in a sentence based on the previous context.
Google BERT: BERT is trained using large amounts of unlabeled text to learn linguistic and contextual patterns. Additionally, it undergoes supervised training using specific tasks to fine-tune its representations and enable classification and information extraction tasks.
Usage and Applications:
Me (ChatGPT): I am used in virtual assistance applications, chatbots, and general natural language guidance. My main focus is on conversation and the generation of contextually relevant responses.
Google BERT: BERT is used in a wide range of natural language processing applications, including text classification, entity tagging, machine translation, and summarization.
In summary, both myself (ChatGPT) and Google BERT are language models based on transformers, but they differ in their objectives and applications. While I am focused on generating responses in conversations, BERT excels in contextual comprehension and is used in various natural language processing tasks.
Google Bard
The first thing that happens when we ask Bard to compare itself with ChatGPT is that it doesn’t understand Spanish, so we had to request it in English, and this is an image of the first draft it showed us.
But we’re not going to be mean and ironically we’re going to use ChatGPT to translate Bard’s comparisons into Spanish (we asked it to respect Bard’s table format).
Conclusion
To wrap up, artificial intelligence (AI) has made significant advancements in recent years, with models like Google Bard and ChatGPT leading the way. These transformer-based language models have demonstrated their ability to understand and generate coherent text, as well as perform specific natural language processing tasks.
However, along with AI advancements, there are also emerging challenges and ethical considerations that need to be addressed. These include data gaps, algorithmic bias, data privacy, and ethical dilemmas in the creation and use of AI. It is crucial to tackle these challenges to ensure the ethical development and application of AI that benefits society as a whole.
Ultimately, AI continues to evolve and present new opportunities and challenges. As we move forward, it is crucial to continue exploring ethical solutions, promoting transparency in the development and use of AI models, and fostering collaboration among different stakeholders to ensure responsible AI use and benefit society as a whole.
For more information on algorithmic bias and ethical dilemmas in AI, I recommend reading this article about Sam Altman (CEO of OpenAI) in Wired, discussing the creation of a regulatory body in the USA for controlled development of artificial intelligence.
