Wednesday, January 18, 2023

Google on AI: Deploying new systems cautiously, like LaMDA and Waymo


Google has shared its perspective on the development of artificial intelligence (AI) in a blog post and in a detailed paper, explaining why it is taking its "AI-based" innovations, such as LaMDA and PaLM, more slowly.

In a blog post and in-depth paper, Google discusses its perspective on the development of artificial intelligence (AI) and explains why it is moving at a slower pace when introducing some of its more recent "AI-based" innovations, such as the contentious LaMDA chatbot and Waymo self-driving car. The blog post is attributed to Sundar Pichai, CEO of both Google and Alphabet, Demis Hassabis, CEO and co-founder of DeepMind, Marian Croak, and Jeff Dean, lead of Google's AI division, as well as James Manyika, SVP at Google.

The "explanation" from Google comes at a time when AI will be the big buzzword of 2023 thanks to the ChatGPT chatbot from OpenAI. It has been dubbed the end of Google's search dominance by many, and reports have also mentioned that ChatGPT has set off alarm bells at the tech giant. Google executives discussed the company's strategy for implementing AI in the blog post, stating that it "is an exciting time in the development of AI."

The blog post continues by stating that Google's technology development will need to address the issues of complexities and risks associated with AI. It adds, "It is critical that we collectively earn public trust if AI is to deliver on its potential for people and society." "We also believe that getting AI right — which to us involves innovating and delivering widely accessible benefits to people and society, while mitigating its risks — must be a collective effort involving us and others."

Meanwhile, Google employs a "scientific method to AI R&D with research rigor, peer review, readiness reviews, and responsible approaches to providing access and the externalization and use of our innovations," according to the paper. Google's delay in releasing some of its most significant AI products to the general public can be seen as an explanation for this.

The paper says that it is "performing continuous adversarial and related forms of testing" and has taken a "differentiated and careful approach" to accessing and deploying novel systems like LaMDA, PaLM, and Waymo. The LaMDA chatbot from Google is based on a large language model (LLM) that is similar to ChatGPT and can have intelligent conversations. An engineer's claim that the chatbot was sentient brought LaMDA into the news last year. Google had also shown how LaMDA was being used to help writers write fiction, even though the chatbot was more of a helper at the moment, as it was acknowledged.

PaLM is another language model that stands for Pathways Language Model (PaLM). It will be larger than LaMDA, support translations into other languages, and GPT-3 (ChatGPT is based on GPT 3.5). However, PaLM has not yet been implemented. The same is true for LaMDA, which users cannot access. The company's AI Kitchen app is the only way to sign up for beta testing. Waymo is a self-driving project that Google is currently testing in the United States.

The paper also gives a detailed list of the difficulties and dangers associated with using AI tools. Additionally, Google emphasized that its core products, such as Android phones, Search, Maps, Photos, and Workspace, already incorporate AI.

Transformers, Word2Vec, Sequence to Sequence Learning, Federated Learning, Model Distillation, Diffusion Models, Deep Reinforcement Learning, Neural Nets with Tree Search, Self-learning Systems, Neural Architecture Search, Autoregressive Models, Networks with External Memory, Large Scale Distributed Deep Networks, Tensor Processing Units are just a few examples of the underlying AI technology that powers other products. For instance, "Transformer" is the technology behind ChatGPT; "GPT" refers to "Generative Pre-trained Transformer."

The conclusion of the blog post reads, "excited about what lies ahead in 2023 and beyond as we get ready to share some new innovative experiences!" Google executives are quoted as saying this. Is it possible that Google will demonstrate new AI products at its annual I/O conference in the coming year? Regarding that, we will need to wait and see.

Catch Daily Highlights In Your Email

* indicates required

Post Top Ad