Google Gone Bard

In the generative AI race,?Google ?is the steady tortoise while Microsoft is the confident hare, poised to unleash the true power of AI at the finish line. Will Google maintain its dominance or will Microsoft emerge as the new king of the hill? Let’s find out.

Google is in a bit of ‘The Innovator’s Dilemma ,’ a term coined by Harvard professor and entrepreneur?Clayton Christensen ?in the 90s. As an incumbent with a profitable search business, Google can’t afford to lose ground. However, with Microsoft partnering with OpenAI, it is potentially losing business to generative AI tools. It’s high time Google started to capitalise on generative AI without hurting its search business.?

As reported by?WSJ , Google currently has over?90% of the search market. Microsoft expects to make?$2 billion?for every?1% increase?in market share. In other words, Microsoft is looking to make a lot of money by?gaining market share in search.

As per Similarweb, Microsoft’s Bing search saw a?25% increase ?in monthly page visits last month, where it stood at?1.4 Bn, compared to?1.1 Bn in February 2023. The increase in its visits shows that its integration with?GPT-4 ?into search is working in?Microsoft’s favour , where it is closely monitoring and increasing the prompt requests per user, alongside improving the model efficiency, user experience, and search results.?

Google , on the other hand, has been trying really hard to make generative AI work for them as well as its customers, responsibly. But, the?response has been quite poor so far. ?Last month, the company decided to give access to?Bard ?– a?ChatGPT ?rival – to users in the UK and the US. The reaction among the public to it has been pretty lukewarm, and?nothing close to ChatGPT .?

“It's not surprising to me, that's the reaction,” said Google chief?Sundar Pichai , in an?interview ?with The New York Times, saying that they have put out smaller models, where they are using a lightweight and efficient version of LaMDA. Looks like Google is still recovering from the mishaps it created during the?initial launch ?of?Bard , which?cost them a lot of money .?

Pichai said that at the beginning of a journey they have to get it right by putting out accurate models, gathering user feedback cycles, improving models, and building trust and safety measures. This is really crucial for?Google ,?

Sadly, a lot of shots have been misfired by Google already, and the more troubling aspect is that a lot of its researchers are?joining its arch-rival OpenAI , and some are even mocking Google for?training Bard on ChatGPT data . However, Google has?denied these claims .?

Amid all the chaos and backfiring, Pichai remains calm and optimistic and said that in the coming weeks, they would be integrating more?powerful models to bring in more capabilities . It is even thinking of adding conversational AI features to its flagship search engine, alongside bringing personalised?Bard to its Google Assistant ?and more.?

For that, we’ll have to wait until the?Google I/O ?conference, which has been scheduled on May 10, 2023. Hopefully, Google will pull this off smoothly.

Top Stories of the Week >>

Quantum Computing is a Failed Cause?

According to a lot of experts in and outside the field,?quantum computing ?is on the brink of a breakthrough similar to that of AI. However, reality paints a vivid picture. A recent study has challenged one of the primary algorithms used to?demonstrate quantum computing's potential , suggesting that it does not provide practical speed-ups compared to classical systems.?

"Mathematicians no longer have proof that quantum computers will outperform classical ones, and this will likely remain the?case for decades ," explained?Xavier Waintal , one of the authors of the study. "While there may be some problems where quantum computing is faster, we have no reason to believe it will be advantageous for most cases."?Read the full story here.? ?

OpenAI’s Version of AI Safety is More Fluff Than Real

With the rising?mutiny ?against the rapid growth of generative AI,?OpenAI’s half-baked blog ?on addressing?AI safety ?couldn’t have been timed any better.?

However, the company’s statement only scratched the surface without addressing?real problems ?that are emerging with ChatGPT and GPT-4. The company sticks to the principle of “building increasingly safe AI systems” without actually talking about the “how” part.?Read more here.

People & Technology >>

Ex-Intel Exec Raja Koduri Gives a Sneak Peek into his New Startup

Last month, Raja Koduri, the man driving Intel’s revolutionary CPU, GPU and AI architecture efforts announced his resignation from Intel Corporation. That too three months after being appointed as the chief architect. This announcement comes at a crucial time for Intel as it prepares to?face off against its primary competitor, NVIDIA.?

This move by Koduri has caught many industry leaders off guard, as he was given a significant position at Intel merely months before his resignation. To gain insight into his decision and his new venture,?AIM?got in touch with him.?

Learn more about Koduri’s future plans?here .

AIM Shots >>

Meta recently released?SAM—Segment Anything model. The model aims to segment parts of an image and detect objects it has never seen before.?Read more here .?

Google announced a preview of?AlloyDB Omni, ?a?self-managed and downloadable version of?AlloyDB ?which can be run by users on-premises, on their edge devices and even on their developer systems.?

OpenAI ?is currently facing?legal ?issues in several jurisdictions. For instance, in Australia,?Brian Wood , Mayor of Hepburn Shire, is preparing to sue OpenAI if ChatGPT’s erroneous statements about him serving a prison term for bribery are not rectified.?More details here. ?

Exotel recently launched its new product, ‘ExoMind ’—a user-friendly,?no-code tool?enabling enterprises to create their own advanced chatbots within minutes.?

UC Berkeley researchers have released a dialogue model based on Meta’s LlaMA for research purposes named?Koala.?Check out the details of the model here.?

A team from UC Berkeley, CMU, Stanford, and UC San Diego has developed?Vicuna-13B , an?open-source alternative to GPT-4, which reportedly achieves 90% of ChatGPT’s quality, and the cost of training the model was around $300.?More details here.?

The much-awaited?Django 4.2, designated as?long-term support ?(LTS), has finally been released.?Check out the features and capabilities here .

The latest version of Pandas –?Pandas 2.0?– was also released this week.?Check out the latest features and capabilities here .?

Google recently introduced ‘ViT-22B’ by scaling?vision transformers ?to?22 billion parameters —which is 5.5 x larger than the previous vision backbone?ViT-e ?which had 4 billion parameters.?More details here.?

At the ‘Minerva Dialogues,’?Pope Francis?officially called ?for the ethical and responsible use of AI.?Read more here.