Reply to this post
This could be a repeat of VHS vs. Betamax.
- Google is adding functionality to allow Google Assistant to compete more directly with Amazon’s Alexa, but what it really needs is to offer love and support to developers of smart home products.
- Google’s failure to do this was visible on every stand at CES where a smart home product was to be found as they all will work with Amazon Alexa
- Only a very tiny fraction will work with Google Assistant.
- Google’s shopping functionality has involved singing a up a series of retailers such as Costco, PetSmart and Target to link their online ordering systems with Google Home such that a similar (to Amazon) shopping experience can be offered through the device.
- Measuring up to Amazon in this category is going to be tough because Amazon has one system through which millions of products are available globally, whereas Google will have to sign up lots of retailers in every locality where it aims to have this service available.
- However, when it comes to almost all of the other features, Google Assistant is capable of offering a vastly superior user performance than Amazon Alexa.
- This is because the AI that powers Google Assistant is top of the class while Alexa’s is second rate at best.
- Furthermore, the Google Home speaker is $50 cheaper than the Amazon Echo and in my opinion, a nicer looking product.
- However, where Google falls over is home automation and here Amazon is currently ruling the roost.
- RFM research has found that device developers receive plenty of love and support from Amazon which combined with the fact that there are now 8m devices in the hands of users drives them to make their products work with Alexa right from launch.
- This is despite the fact that using many of these products with Amazon Alexa is a frustrating and fragmented experience.
- A good example of this is Plex, which recently enabled an Alexa skill so that the user could control the Plex player using Alexa.
- However, because Alexa lacks the brains to make service intuitive, the user experience is so bad that one tries to control Plex with Alexa once and quickly returns to the remote control.
- In contrast to Amazon, many developers find that Google is difficult to work with and some did not even know who to at Google to call to enable Google Home with their product.
- This is the opportunity for Google Home even though it only has around 0.5m devices in the market today.
- I think Google needs to ramp up its love and support for developers immediately and thinking that they will just turn up at Google i/o is not nearly good enough.
- There is a whole segment (home) of the digital ecosystem up for grabs right now and I still maintain that this is Google’s to lose.
- However, at the moment it is Amazon that is blazing the trail and if Alexa makes it into the majority of households before Google pulls its finger out then the game will, in all probability, already be lost.
- This will not be the first time that an inferior product will have won the day and I think there are valuable lessons that Google can learn from studying this history.
- From an investment perspective, I continue to not really like either Alphabet or Amazon preferring Baidu, Tencent and Microsoft.
Reply to this post
I think Huawei would be better off doing a deal with Baidu.
- It looks like Huawei has decided to build its own Chinese language digital assistant to cement its recent gains at home but no matter how many bodies it throws at this task, its lack of the core raw materials (data and history) is going to cause problems.
- The digital assistant is the first real Digital Life service that is entirely dependent on artificial intelligence for its functionality which creates a huge challenge.
- Furthermore, in order to evolve, all digital assistants need to generate usage data which can then be used to improve the algorithms that power the user experience.
- Even the best assistants out there today are hugely limited in terms of what they can understand and what they can achieve.
- For example, to accurately answer questions around exchange rates, the assistant has to be taught what these are, how they work and in what form the questions are likely to be asked.
- For example, asking Amazon Alexa how many US Dollars there are to the GB Pound provides the correct answer but ask for UAE Dirhams to the Pound or Dollar and Alexa falls silent.
- Only Google Assistant was able to provide the right answer due to the combination of the best search system and the best AI available.
- In effect RFM research has found that Alexa, Cortana and Siri have been programmed with a fairly narrow set of capabilities and the AI and data set is simply not there to support the service when something unexpected is requested.
- Fortunately for Huawei, Google is not present in China but at home it will be facing an opponent that is almost as good: Baidu.
- Baidu dominates the search market in China and has been working on its AI algorithms for nearly 20 years.
- Furthermore, Baidu has already launched its own digital assistant called Duer which I suspect will be significantly better than anything that Huawei is likely to produce in the medium term.
- However in China, none of the ecosystems are preinstalled devices meaning that Baidu will be unable to install Duer on the device and set it as default.
- RFM research (see here) has found that this could confer a substantial advantage to any ecosystem as strategy is virtually absent in the Chinese market outside of the app stores.
- Huawei as a handset maker will have this advantage and so I can see a scenario where users try its digital assistant but unless it is superb they will quickly switch to Duer.
- This is where I think Huawei will have difficulties as even though it has 100 engineers working on this product, it is starting from scratch and building decent AI takes years and requires vast quantities of data.
- Hence, I think it unlikely that Huawei will ever come up with a product as good as Baidu’s.
- This is where I think Huawei and Baidu could help each other as Baidu has the product and Huawei a mechanism for distributing it.
- A deal where Huawei installs Duer at the factory and sets it by default in return for being paid TAC (traffic acquisition cost) makes more sense to me than paying 100 engineers to come up with an inferior product.
- This will not help Huawei’s ambitions to develop an ecosystem and generate better profitability, but TAC revenue from Baidu would certainly help improve margins.
- Given its recent market share gains at home, the time to negotiate this deal is now rather than when its own assistant has tried and failed.
- Although Baidu looks like it may be backing out of its ecosystem, the short-term improvement in its financials that cost cuts could generate could give the shares a lift (see here).
- This is why it is still on my preferred list along with Tencent and Microsoft.
Reply to this post
Focusing on video first makes complete sense.
- I think that Facebook is making the right choice in targeting video first as it already has traction and video-based services tend to have the lowest requirements for artificial intelligence to make them easy, fun and useful.
- With the launch of a TV app being just the latest move Facebook has made in video, it is increasingly clear that Media Consumption is Facebook’s number 1 priority for 2017.
- The TV app that is being launched is very simple in that it makes it easy for a user that does not have time to watch videos on Facebook during the day to easily to so at night on a larger screen.
- This should enable a better video experience and begin to spread engagement across other devices but it will come with the added complication of multiple resolutions and bit rates.
- On a mobile device the screen is small which means that lower resolution videos and bit rates are acceptable, but once these are played on a larger screen, their shortcomings quickly become obvious.
- This move into TV comes hot on the heels of the addition of a tab at the bottom of the Facebook app which links to the top trending videos as well as videos that Facebook thinks that the user might like.
- The TV app will initially be available on Amazon TV and Apple TV but I expect that it will quickly spread to Xbox, PlayStation and the other streaming TV devices that are available.
- The one place I don’t expect to find it is Chromecast as Facebook’s video aspirations are clearly a challenge to YouTube.
- Of the three new areas of Digital Life (Gaming, Media Consumption and Search) that I see Facebook targeting (see here), going for video first makes complete sense.
- This is because Facebook already has a lot of traction in this space and also because it is the least demanding in terms of requiring intelligent automation.
- The total number of video items that are present is very low compared to other things like music or searches and knowing who posted the video is a good indicator of its content and who will like it.
- I continue to see Facebook as the laggard in AI (see here) and targeting video is sensible as it gives it more time to improve its AI before having to apply it to more difficult tasks.
- Furthermore, the fact that video is a fast growing, but likely soon to mature, medium for digital advertising also means that the time to really address it is now.
- I see the app on the TV as just the beginning and would not be surprised to see this being followed up with premium content taking it into the realm of Netflix, Hulu, YouTube and Amazon Prime.
- That being said, I don’t think that Facebook’s offering in Media Consumption is anything like mature and so I think it will be some time yet before it becomes a real destination like YouTube.
- Consequently, I still see a slow period of revenue expansion while its new strategies mature before revenues take off again.
- As this reality sinks in, I think the valuation could unwind somewhat providing a better opportunity than now to invest for the long-term.
Reply to this post
Facebook and Alphabet are worlds apart when it comes to AI.
- While Google is pushing the boundaries of artificial intelligence forward, Facebook is making excuses for its inability to control hate speech highlighting once again who is the leader and who the laggard in the field of AI.
- Google has demonstrated a system that it has built that allows very poor digital photos to be enhanced to reveal details of the photograph that have been completely obliterated.
- In a similar fashion to the way that Deep Mind built AlphaGo, Google has combined two neural networks to produce the algorithm capable of enhancing low resolution images.
- One of these networks uses its knowledge of certain images to add details while the other effectively reverse engineers the process by which the image was compressed into its current form.
- The result is quite startling but it is worth remembering that the machine knew what the original image was a face or a bedroom but no more than that.
- Despite Google’s claim that this was an experiment only with no plans to put it to use, I think that the uses for this are endless.
- This technology would be useful in upscaling video to high resolution screens as well as being highly applicable to law enforcement, security, military, medical and so and so forth.
- Hence, I think that this technology or an off shoot of it is likely to find its way into Google’s products and services in the medium term.
- To me this is another demonstration of how well Google leads the field of artificial intelligence and is the closest to using it to enhance the richness and quality of its Digital Life services.
- This will be a huge benefit to Google as better services will drive more usage through its networks giving it a greater opportunity to monetise.
- However, this is also the opportunity that Facebook is chasing but when it comes to making its Digital Life services deeper and richer with intelligence, I see it being miles behind.
- The problems that it has had with fake news, idiotic bots and Facebook M, all support my view that when Facebook tries to automate its systems, things always go wrong.
- The problem is not that Facebook does not have the right people but simply that it has not been working on artificial intelligence for nearly long enough.
- RFM research has found that time is the single most important element when it comes to having a solid foundation of intelligent algorithms upon which to build intelligent services (see here).
- In contrast, Google has been working on this for over 20 years and is still innovating as fast as it can.
- Facebook’s most recent pronouncement by one of its lawyers that it is unable to control hate speech on its platform due to the scale of data that is posted every day, is just another data point highlighting the problem.
- Facebook has to get this under control otherwise I fear that it will fail to really expand beyond social networking and instant messaging as the offerings from its rivals will be more useful and more fun.
- Facebook has some time to get to grips with this problem but I still think it will have to resort to making a series of acquisitions in order to catch up with its rivals.
- I remain uninterested in both Alphabet and Facebook at this time preferring Microsoft, Tencent and Baidu with Apple for income based investors.
Reply to this post
Something must-have has to come out of the bag at launch.
- Essential Products Inc. is the latest in a long line of protagonists which is aiming to have a crack at the high end of the handset market and while it knows where it should compete, I am not convinced that it will be able to.
- Essential is a start-up led by Andy Rubin who was the founder of the company that Google very successfully turned into Android.
- Essential aims to compete in high-end consumer electronics by offering differentiation if the following areas:
- First: Hardware. In any ecosystem strategy today, the most important device is the smartphone as this is where users spend almost all of their time.
- Essential aims to compete here by offering a device which has a screen with no or almost no bezel.
- Xiaomi has already done this quite effectively with the Mi Mix but this currently only available in China.
- This is also the strategy, that I think Apple might use in the iPhone 8 but Essential should be able to launch well ahead of this.
- Second: Artificial Intelligence. RFM research has recently identified AI as the 8th Law of Robotics concluding that AI is likely to have meaningful impact on the quality, and hence appeal, of Digital Life services in the medium term.
- However, RFM research has also concluded that good AI requires a huge amount of time and a vast trove of user data in order to develop.
- I seriously doubt whether Essential has either of these characteristics and while it may try to develop intelligent services, I suspect that it will struggle.
- This is especially the case as its services are likely to end up competing with Google’s own which Essential will be obliged to implement on its smartphone and to set as default.
- Third: Mods. Essential’s patent filings include a design for a proprietary magnetic charging port that can also be used as an expansion slot to add hardware functionality to devices.
- In this day and age, unless you are Apple, proprietary charging ports are a big no-no and may be an indicator of the inexperience that Essential has in making smartphones.
- Furthermore, the best mod on the market is currently made by Motorola which already has some volume and a reasonable range of third party brands making devices to connect to it.
- All other modularity plays have either already failed or are struggling for relevance.
- I struggle to see how Essential is much different.
- Fourth: Cross device: Essential intends to produce a series of devices that will work together to deliver its proposition to users in all aspects of their Digital Lives.
- This is also a vision pursued by Samsung, Xiaomi, Apple, Microsoft among others, and it does make some sense.
- This is because if devices from one manufacturer all work together seamlessly, it provides a reason for users to buy all of their other devices from the same manufacturer.
- This is one way of generating device preference without having an ecosystem and hence of earning better than commodity returns.
- The problem is that getting all of these devices to work seamless together is fiendishly difficult and even the mighty Apple has not really got it right.
- Microsoft does a reasonable job but there are still glaring holes in the experience that it offers.
- Combined with this difficulty, will come the capital intensity of having to design a series of device types all at the same time which could easily be beyond the financial resources of Essential Products Inc.
- With the segment that Essential chosen to compete in, there is very little scope to compete in the ecosystem as it will invariably have to support Google.
- This is because its target users whether they are on iOS or Android will already have a meaningful part of their Digital Lives invested in Google are unlikely to want to switch.
- Consequently, I remain uncertain as to what is special or different with regards to Essential Products Inc.
- I am hoping to be well informed when it launches its proposition at some point during H1 2017.
Reply to this post
December 16th 2016: Radio Free Mobile updates its flagship research product with the publication of: Mobile Ecosystems – Artificial Intelligence – Men and boys.
RFM research subscribers will receive their copy directly by email.
Click here for a sample and here for purchase options
The difference between men and boys will be the brains of their toys. Artificial Intelligence promises to substantially improve the Digital Life services offered by the ecosystems which has underpinned a period of feverish investment. Despite this activity, developments are at a very early stage with none of the big challenges of AI being close to being solved. It is the search engines that are ahead in AI followed by Apple, Microsoft and Amazon. AI remains the Achilles heel of Facebook.
- Artificial Intelligence appears at last to be coming of age. The prospect of making real returns on investment has driven all of the ecosystems to invest heavily.
- Three goals for AI. AI is still in its infancy with three big issues to be solved. These are: 1) the ability to train AIs using much less data than today, 2) the creation of an AI that can take what it has learned from one task and apply it to another and 3) the creation of AI that can build its own models rather than relying on humans to do it. Performance in solving these three problems is likely to separate the men from the boys in the long-term.
- Early days. RFM finds that most claims to AI are simply advanced statistics and that true AI is at a very early stage. Even the best have made little headway with the three goals of AI.
- Law of Robotics. There is no doubt that good quality AI has the potential to significantly improve the quality of Digital Life services offered by the different ecosystems. Consequently, RFM sees AI being a major differentiator and now includes an assessment of AI as Law of Robotics No. 8: An ecosystem must have good artificial intelligence.
- Digital Assistants are the first real deployment of AI in ecosystems and are being offered free in order to generate the data that is needed to continually make them better. Consequently, digital assistants are a good first yardstick of each ecosystem’s competence in AI.
- Search engines. AI still requires vast amounts of human labour, great skill and copious data to develop which hands a substantial advantage to those that have been doing it the longest. Understanding data has been the livelihood of the search engines for many years. This is the main reason why it is Google, Baidu and Yandex that are the global leaders in AI and all of them are aggressively investing to maintain their advantage.
- Fast followers are made up by Microsoft, Apple and Amazon. Both Microsoft and Amazon have scope to earn a return on AI in their businesses that are not part of the ecosystem. Apple appears to have voluntarily hobbled its AI development with differential privacy.
- Facebook is the laggard with one of the weakest positions in AI globally. RFM research indicates that Facebook has real problems with automation. These have to be fixed otherwise providing customised services to 1.8bn users manually will be cripplingly expensive
Reply to this post
Amazon is stepping up its efforts in the coming battle of the brains.
- Amazon is following its competitors by announcing a series of services based on its AI research but the performance of Echo makes me think that there is still an awful lot of work to do.
- The headline of the AWS re:invent conference was the launch of three new services based on the AI work that it has done to date.
- First: Amazon Recoknition.
- This is a service that provides image and facial recognition as well as being able to assess the mood of the subject as well as detect glasses and so on.
- Second: Amazon Polly.
- This is a text to speech service that takes text input and returns an MP3 stream that sounds like a real conversation rather than a mechanical expression of just words.
- Amazon Lex
- This is a service that provides natural language understanding and automatic speech recognition.
- It is also the system that powers the speech recognition of the Alexa digital assistant in this regard it performs reasonably well.
- It is in Alexa’s ability to understand context, multi part enquiries and respond to enquiries that it falls short.
- These launches have also gone hand in hand with a new initiative with Intel where Intel will launch a reference design for a home speaker that includes the Alexa digital assistant.
- This is a good idea because at this early stage all the digital assistants need to collect as much data as possible in order to improve.
- Hence, I expect that there will be a slew of devices that use Alexa for their brain with a range of designs, capabilities and price points.
- I would not be surprised to see Google follow suit in this strategy with Google Assistant where, once again (see here), it will have the advantage.
- This is because I have long believed that Google Assistant is vastly superior to anything else in terms if the user experience that it offers and the usefulness of its responses.
- Consequently, if Google executes well and users like its product, it should be able to encourage manufacturers to deploy its digital assistant rather than Amazon’s.
- Once again Google’s success (just like Pixel) is hinging on execution and only time will tell whether the recent re-organisation has been successful in solving Google’s historical problems in this area.
- I continue to think that Alphabet is pricing in success in most of its endeavours and so prefer Microsoft, Tencent and Baidu from an investment perspective.
Reply to this post
Inability to automate bites Facebook again.
- Facebook’s problems with fake news once again demonstrates the real problems that this company has with anything complex that is not carried out by a human.
- During the recent presidential election, it became clear that a large number of users rely on Facebook, Twitter etc. for current affairs rather than the traditional news broadcasters.
- This puts a huge responsibility on these players to only disseminate stories which are factually accurate which is something that Facebook in particular has been struggling with.
- Facebook has reacted quickly to address this issue but the remedies that it is putting in place reveal once again that Facebook relies almost completely on humans to carry out intelligent tasks.
- Facebook’s remedies to address fake news include:
- First: Use third party organisations as fact checkers
- Second: Verifying information with journalists.
- Third: Relying on the community to report fake news.
- Very much like Facebook M and Facebook’s bots, this remedy relies almost exclusively on humans to solve difficult problems, highlighting yet again how weak Facebook is when it comes to artificial Intelligence (AI).
- This is a problem because humans are extremely expensive compared to machines and with the high level of traffic and content growth that Facebook is experiencing, quality checking is already a massive task and will only become greater.
- Facebook is trying very hard in AI and is recruiting as fast as it can, but AI takes a very long time to get right and I am concerned that Facebook will end up with a much higher cost structure than its rivals.
- This will mean that it will have a higher cost base than Google or Baidu leading to lower profitability.
- Given Facebook’s late entry into AI, I think that it will have to buy its way out of these problems because growing it organically is likely to take much too much time.
- Although this is a very serious problem, I have some confidence that Facebook will be able to sort it out.
- This confidence is based on the exceptional execution and implementation of a system for monetisation of mobile that led to mobile becoming the majority of Facebook’s revenues in a very short period of time.
- This problem is much more difficult to solve but I am hopeful that Facebook’s exceptional record in execution will see it through.
- This is yet another reason to be cautious on Facebook in the short-term (see here) but assuming that this is fixed, I continue to look for an entry point into the shares during 2017 or 2018.
- In the meantime, my preference remains Microsoft, Baidu and Tencent.