Bard, Google Assistant to Merge
Google's tentative rollout of Bard hurt the perception of Google's technology leadership. So CEO Sundar Pichai is now doing PR interviews. He spoke recently to NYT podcasters about Bard, AI and ChatGPT competition. Pichai basically said Google was being cautious but that the company had "more capable" language models (vs. competitors) and that improvements would be coming fast. Bard will shortly get better at math and logic and improve across wide range of functions: reading comprehension, general knowledge translation, content summarizing and so on. Google is also merging the Bard and Google Assistant teams. Google Assistant launched in 2016 to great fanfare. It was positioned as a personalized tool that could search but was more than a search engine. It could help you "get things done." Google Assistant – as well as Alexa, Cortana and Siri – mostly failed to live up to the hype. The integration of Bard and Google Assistant could help the company deliver on its original promise.
- Bard and Google Assistant could remain separate "products." But Google may be forced to choose a consumer brand in the near future.
- Of all current competitors, Google is best positioned to realize the virtual assistant vision. Nobody else can match their mix of assets.
- The (personalized) assistant is a successor to search. Google said as much in 2016. But monetization remains a major challenge.
GM Boots Apple, Partners with Google
The rumor that Apple is building a car goes back several years. The project ("Titan") has been plagued by turnover and delays; and it's not clear whether an actual Apple car will eventually materialize. But last year, at its annual developer conference, Apple showcased a dramatically expanded CarPlay that looked more like "CarOS" than an infotainment system. This unsettled some of the major automakers who have generally benefitted from the inclusion of CarPlay. Some may now see an expanded CarPlay as an obstacle, as they seek to turn their cars into rolling smartphones. Indeed, car makers hope to exploit user data and become data brokers. In that spirit, GM has decided to phase out CarPlay (and Android Auto) in its EVs starting with the 2024 Chevrolet Blazer. GM will instead work with Google to build a native infotainment system, allowing GM to (mostly) own the CX and customer data. Google aims to replicate this strategy with others, although broader demand is unclear.
- This makes abstract sense for GM, which wants to build an "integrated" CX. Google is seen as bringing more to the table than Apple.
- Regardless of GM's control, Google will get data access. GM's homegrown infotainment is mostly a failure; Google gives it a better shot.
- On some whiteboard somewhere is a Google-GM ad strategy, including localized ads – with a revenue share for GM.
BloombergGPT: An Enterprise Model
Last Thursday Bloomberg announced its own AI-GPT offering: BloombergGPT. The big news is that it's trained on proprietary financial data. Bloomberg said internal teams drew upon its "extensive archive of financial data to create a comprehensive 363 billion token dataset consisting of English financial documents ... augmented with a 345 billion token public dataset to create a large training corpus with over 700 billion tokens." The company added, "The BloombergGPT model outperforms existing open models of a similar size on financial tasks by large margins." This combination of proprietary and public data will become a model for enterprises going forward. (It's the model for ChatGPT plug-ins.) All major software and data vendors will need to provide some AI-augmented capability to customers; competition will demand it. Extensive, high-quality data will be a major differentiator going forward.
- One question in the rapidly evolving AI ecosystem is: where's the value creation? It could be the LLM (e.g., OpenAI, Google, Claude).
- But the LLM and basic generative AI functionality could also become commoditized. It's not clear yet.
- Value will ultimately reside in some mix of proprietary data, brand, pricing and UX. Incumbents that act quickly will potentially avoid disruption.
- Near Memo ep. 107: Local Search Ranking Factors with Darren Shaw
- Local Search Ranking Factors 2023, by Greg Sterling
- Google ML can identify spam phone numbers in local listings.
- BingGPT improves local results and more.
- Ad placements in AI results: fewer slots, higher costs.
- Opinion: AI will become a new channel; don't "robots.txt" it.
- AI ethics leaders: don't focus on utopia or apocalypse scenarios.
- Wix integrates with Meta to manage interactions with Wix CRM tool.
- FTC readies case against Amazon (Alexa) for data collection on kids.
- Amazon opens up its local Sidewalk network for IoT developers.
- Petty: Twitter unverifies the NYT after they refuse to pay for Blue.
- Select Twitter VIPs get extra visibility with company's algorithm.
- Most Americans unhappy with jobs, but half don't take available PTO.
- Parisians overwhelmingly vote to ban scooters (Lime, etc.).
- Content quality for SEO and beyond: what to measure and how.
Listen to our latest podcast.
How can we make this better? Email us with suggestions and recommendations.