Home » Technology » Google » Google introduces new PaLM 2 AI language model

Google introduces new PaLM 2 AI language model

Google

After the integration of ChatGPT into various services of Microsoft, both companies are trying hard to come up with something exceptional in the conversational AI field i.e., GPT-4. In order to compete with Microsoft and OpenAI, Google has come forward with a new state-of-the-art AI language model termed PaLM 2. As per recent information, the language model is trained for various tasks like math, coding, reasoning, multilingual translation, and natural language generation.

In accordance with the senior research director at Google, the language model has been trained in 100 different languages. Given this fact, PaLM 2 can easily understand idioms and phrases in different languages. Additionally, it can also provide better reasoning for different queries. Well, this is better since language models can portray false information that sounds real.  

Besides this, the tech company has developed different versions of PaLM 2 that cater to the specific requirements of enterprise customers. One such example is Med-PaLM 2. Reportedly, it is trained on health data. It can answer queries just like an expert answer. Similarly, another version named Sec-PaLM 2 helps to detect threats in code. It can even explain the behavior of malicious scripts.

The tech company Google claims that PaLM 2 is already utilized in 25 different features and products. It includes Bard and Google Workspace apps like Docs, Slides, and Sheets. Furthermore, Google has also created a PaLM 2 version for smartphones termed Gecko. It has the capacity to process 20 tokens per second.

The new PaLM 2 language model by Google is a sophisticated one. The company has started working on its second version. it has been named Gemini. Reports suggest that it will be more efficient and multimodal. With the advancements in AI systems, threats and concerns are also sparking. Companies are required to observe strict measures to make sure that this fast-paced development of AI systems doesn’t impart lethal impacts.