All posts by Jim Muttram

The challenge of AI in the office


The evolution of technology has always had profound effects on how business is done. The integration of AI into office software is no exception. As we inch closer to a future where AI becomes a dominant force in our work environment there are consequences good and bad which will undoubtedly emerge. Here are some early thoughts:

The Development of Online Meetings

The convenience of online meetings is undeniable, and the integration of AI only amplifies their advantages. With features such as real-time transcription, automatic summaries, and the creation of meeting notes and action items, AI tools are making online discussions more productive and accessible. It may well be that attending online, now mostly seen as the inferior option, actually become preferable as they new capabilities add real and attractive functionality. The challenge for those whose businesses who want an effective blend of in-person and online meetings will be replicate those benefits in the meeting room as well over the Internet.

The Need for Speed vs. Quality of Decision Making

Integrating AI into office tools speeds up various processes: information is sorted, processed, and presented faster than ever. This can lead businesses to make decisions quicker, riding on the momentum that technology provides.
However, with this increased pace, there’s a danger of reduced reflection time, making decisions which have not been given ample consideration more likely. This will almost certainly lead to poorer outcomes. And the ease with which AI can process and present data might push organisations to act impulsively, with unpredictable, and probably undesirable, results.

Spurious Professionalism

The use of sophisticated AI tools can lend a, perhaps unwarranted, aura of professionalism to presentations and decisions and there’s a real risk that flawed logic or inaccurate data might be glossed over simply because it’s presented in a polished, AI-enhanced manner. We are already biased to see form winning out over content and this is likely to get worse, quickly.
A tool is only as good as its user, and even the most advanced AI can’t make up for a lack of proper review or critical thinking. Relying heavily on AI-driven tools without thorough human oversight can lead to a kind of spurious professionalism—where things look impeccable on the surface but are fundamentally flawed underneath.

In Conclusion…

As AI continues to integrate with office software, businesses will be presented with huge opportunities to improve efficiency and functionality. And competitive pressure will ensure they feel a strong desire to do so. How much real benefit businesses get will depend on thoughtful integration and proper oversight.

I asked ChatGPT to list 10 good and bad things about LLMs

Here are 10 ways in which large language models such as myself could provide real benefit to the world:

  1. Facilitating communication: Large language models can be used to bridge language barriers and facilitate communication between people who speak different languages.
  2. Improving healthcare: Language models can help healthcare professionals diagnose and treat patients more accurately and efficiently by analyzing medical data and providing insights.
  3. Enhancing education: Large language models can be used to create interactive and engaging educational materials, such as online courses and textbooks, that adapt to the needs of individual learners.
  4. Boosting scientific research: Language models can assist researchers in analyzing large datasets and identifying patterns and relationships that would be difficult to detect manually.
  5. Improving customer service: Language models can be used to provide 24/7 customer service and support, improving customer satisfaction and reducing wait times.
  6. Enhancing creativity: Language models can assist in generating creative works such as music, art, and literature, allowing for new forms of expression and innovation.
  7. Advancing business: Language models can be used to automate tasks such as data analysis and customer outreach, reducing costs and improving efficiency.
  8. Enabling accessibility: Language models can be used to create more accessible interfaces and products for people with disabilities, including those with visual or hearing impairments.
  9. Facilitating disaster response: Language models can be used to analyze social media data and news reports to provide real-time updates and insights during natural disasters and crises.
  10. Improving personalization: Language models can help businesses and individuals personalize their products and services to meet the unique needs and preferences of their customers.

Here are 10 ways in which large language models could have a detrimental effect:

  1. Bias and discrimination: Language models can perpetuate existing biases and discrimination by reflecting the biases present in the data they are trained on.
  2. Misinformation and fake news: Language models can be used to generate false or misleading information that can spread quickly and easily online.
  3. Privacy and security: Language models can be used to analyze and mine personal data, potentially leading to breaches of privacy and security.
  4. Job displacement: Language models and other forms of AI automation can replace human jobs, leading to economic disruption and social inequality.
  5. Dependence and over-reliance: Dependence on language models and other forms of AI can lead to over-reliance on technology and a lack of critical thinking skills.
  6. Lack of transparency: The algorithms used to train and operate language models can be opaque, leading to a lack of transparency and accountability.
  7. Legal and ethical issues: The use of language models in sensitive areas such as law enforcement and national security can raise legal and ethical concerns.
  8. Loss of human touch: Overreliance on language models and other forms of AI can lead to a loss of human interaction and empathy.
  9. Addiction and social isolation: The use of language models and other forms of technology can contribute to addiction and social isolation.
  10. Environmental impact: The energy consumption and carbon footprint associated with the training and operation of language models and other forms of AI can have a negative impact on the environment.

Interesting……

Elon Musk’s strengths and weaknesses

I’ve been thinking about the melt-down which is happening now Elon Musk was reluctantly successful in his purchase of Twitter.

To kill two birds with one stone I thought I would ask ChatGPT what it thought were his strengths and weaknesses:

Elon Musk is a highly successful entrepreneur and CEO, known for his innovative approaches and ambitious projects. However, like all individuals, he also has both strengths and weaknesses.

One of Musk’s greatest strengths is his ability to think big and envision the future. He has a strong vision for what he wants to achieve, and he is not afraid to take risks in pursuit of his goals. This has allowed him to successfully co-found and lead companies such as PayPal, Tesla, and SpaceX, which have all had a significant impact on their respective industries.

Another strength of Musk’s is his ability to adapt and innovate. He is constantly looking for ways to improve and evolve his companies, and he is not afraid to pivot or change course if necessary. This has allowed him to stay ahead of the curve and anticipate market trends, which has been key to the success of his businesses.

However, Musk is not without his weaknesses. One of his most notable weaknesses is his tendency to be overconfident and make bold predictions that may not come to fruition. This has led to some setbacks and criticism when his predictions have not been met. Additionally, Musk’s strong focus on achieving his goals can sometimes lead to him being perceived as lacking empathy or consideration for others.

Overall, Elon Musk is a complex and multifaceted individual with both strengths and weaknesses. While his confidence and vision have helped him achieve tremendous success, it is important for him to continue to strive for balance and consider the impact of his actions on others.

It’s not a bad answer, and exceptional if you think it took 10 seconds for an AI to come up with it.

It’s not a sophisticated answer: it ignored, for instance, his tendency to treat human “laws” (say the desirability of free speech) as if they were the laws of physics, which is his greatest flaw and which has been mercilessly exposed by the shit-show that is Musk-owned Twitter.

But prior to his latest display of his character I would say this is a pretty good assessment. I can only imagine the havoc that will be caused by ChatGPT and its ilk in the very near future!

the joys of vertical integration

One of the many benefits of Tesla vaunted vertical integration is well illustrated by the announcement that off-peak charging will be free at Tesla Superchargers in the California over the Thanksgiving holiday period.

Because Tesla owns all its own charging infrastructure it has a comprehensive view of charging patterns and spikes across its network. This allows it to react to congestion which can be a major problem where there is a large concentration of Tesla EVs.

The company has already tried variable pricing during the day to even out spikes and it has even developed its own mobile Superchargers to boost capacity at high pressure sites. It is able to do this because it makes not only its own chargers but also Megapacks – very large lithium storage batteries which is sells through its power business.

Another reason why Tesla is going to be so hard for automakers to beat.

Teslas drive further

No, I’m not talking about range – the perpetual first question in any discussion with non-ev-drivers. I talking about a survey by the RAC, a motoring organisation in the UK, which found that Tesla drivers drive more miles on average than drivers of any other car.

In the first three years of owning a new car, Tesla drivers cover an average of 12,459 miles a year. Meanwhile, Mercedes owners clocked 12,100 miles each year, and Volvo owners averaged 11,578 miles.

This compares to an average of 10,377 miles per year for the average of all cars in their first three years of ownership, according to the Department of Transport.

I can corroborate. I drove an average of 10,000 miles a year in the 10 years I owned a Mercedes E Class. Since I have owned a Tesla Model X I have driven 15,000 in nine months and it would have been more if the Coronavirus hadn’t pretty much put paid to driving.

Electrek, who reported on the RAC survey, concludes: “Electric cars with bigger batteries and faster charging get driven and charged more.”

That is true – but the main thing is they are just so much more fun to drive!

Round One: Coronavirus

I took a picture of London from the top of the North Downs in 2017 after I was so struck by the visibility of the pollution hovering over the city. Last week I stopped again at the same spot and took roughly the same photo.

What is so striking is that after only five weeks of lockdown, the dramatic drop in traffic has had such a noticeable visual effect on the air quality.

TomTom, the navigation company, has provided graphs of various cities around the world showing the change in traffic.

What a difference it would make if we could effect a change like this but without the huge downside of a pandemic.

Some cities such as Milan are already planning to reclaim some of their streets inspired by the experience of the traffic-drop. And given that social distancing is likely to be here to stay for quite some time – at least until widespread vaccines are available, others are bound to follow suit.

Wired reports that many cities around the world have already blocked off city streets to provide more open spaces for people to safely navigate.

We could of course go back to normal after the pandemic is over but as The Economist eloquently illustrated coronavirus is merely Round One; the next battle is the big one.

There have been notable examples of self-less co-operation during the coronavirus challenge, but also many examples of narrow-minded, nationalistic responses following the lead of the catastrophically inadequate President of the United States.

We can only hope the sobering example of fighting a pandemic will create real impetus for change which can create a common will to deal with the biggest global challenge of all. Fingers crossed.

Tesla Service

I’ve taken my Tesla to the Dartford Service Centre today to have my camera “calibrated”.

This is needed because I had to replace the windscreen before Christmas after a stone cracked the glass (for the second time in as many months!)

The windscreen was replaced by Autoglass but they can’t calibrate the camera as they do on other vehicles because, they say, “Tesla won’t let them have the software”.

I suspect the reason for Tesla’s reluctance is because of the advanced autopilot system which relies heavily on the front camera for its correct functioning. Any errors in calibration could potentially cause malfunctioning automatic driving which would be something that Tesla would want to avoid.

The actual process is pretty slick – you book the appointment on the Tesla app and all the communication about it is thereafter done through text message.

When you arrive at the service centre you simply drive through the automatic door and then walk through to the adjacent service department to register. The service staff then take pictures of the car to record any damage (for the obvious reasons) and then ask you to sign to consent to the work.

Then, assuming it’s a short job, you sit in the lounge, drink the free coffee and wait until they tell you you car is ready. (If you’re out and about the app will tell you the car is ready.)

This is the experience in the UK but it is already being refined elsewhere in the world.

In Northern California where there are many, many more Teslas YouTuber All Electric describes his encounter with a fully automatic service experience, which, incidentally he wasn’t too thrilled about.

There you drive into the service bay, sign in via the dedicated iPads and get an Uber credit texted to you phone to take you home. Then the app tells you when the car is ready, you get another Uber credit and you can pick it up.

One of the things that frequently happens in the UK at the service centre is that people forget to turn off the pin-to-drive feature which means the technicians can’t drive the car until it has been unlocked with the unique pin you have set. And this can only be turned off in the car itself, not remotely through the app.

It’s not clear how the Californian service centre handles this, or how it deals with any subsequent claims over incidental damage while the car is being serviced in the absence of agreed photographic evidence.

As Tesla’s rapid growth continues it is inevitable that they have to innovate to try to keep up. We can expect more change in the experience in the future.

Tesla’s charging advantage

I travel most months to the Norfolk coast where our cottage does not have off-street parking and I cannot therefore charge my car overnight. As it’s about 150 miles away I need to charge while I’m there in order to have enough charge to get about and then get home again.

North Norfolk is practically a desert when it comes to rapid chargers but an enterprising local business did install one on their forecourt in a town about 15 miles from our cottage and this is where I’ve been charging up to now. (This was the catalyst which allowed me to buy my Tesla in the first place.)

Polar charger at Kellys Motor Services in North Walsham

On my most recent visit I noticed that the Morrisons in Cromer, the local town, has also installed one which is quite a bit more convenient.

Genie-Point chargers now at Morrisons stores

In theory both these chargers offer the same thing. They have three types of connection – AC type 2, Chademo and CCS. The fastest are the last two, both DC connections, and they are supposed to provide up to 50 KW of electricity.

Tesla Chademo adapter on the right and a CCS adapter on the left

In practice, though, not all chargers are created equal. In my my limited experience so far with these two, the Polar charger has provided the best speed using both my Chademo and CCS adapters – about 45 KW on a sustained basis. The Genie-Point charger would only charge on Chademo and gave me 30 KW.

There are differences in how the services work, too. Polar have a free smart card which allows touch access to the charger (you are billed monthly – £7 for the service charge and then electricity which is about as cheap as home). Genie-Point have a web app which you have to have an account which you have to put £10 into, and then it’s pay as you go from there. If you are in an area with poor mobile signal you could be in trouble as the app is the only way I could see of starting a charge, whereas Polar does everything through the RFID card.

So on paper the Polar charger wins, but there is another variable – location. The Polar charger is much further away and is on an out-of-the-way industrial estate which is 15 minutes walk from the nearest coffee shops in town. The Genie-Point charger is in Morrisons’ car park, near their facilities and also five minutes walk from the town centre. If you are charging for 40 minutes which is pretty typical on a rapid charger for a 100 KW Tesla then things like that matter.

Which brings me to the alternative. On my way back to London I have the option to use the Supercharger at Elevdon which is just off the A11.

The eight stall, 150 KW Elvedon supercharger

The eight stalls are never all occupied in my experience and I can get between 80 KW and 140 KW depending on the state of charge of my battery (the lower the better). The supercharger is right next door to a pub and restaurant. This is one of over 70 superchargers in the UK – by far the largest network of fast chargers in the country. To use them I just plug in and off it goes. I get charged automatically on my registered credit card and the Tesla app tells me when I’ve charged enough to get home.

The best option for me then at the moment is to charge in Cromer just enough to get to Thetford and then use the superior Tesla chargers and nearby facilities.

The contrast between my charging experiences in Norfolk demonstrates quite graphically how much of a lead Tesla has in the electric car market, and what a real challenge the other car companies who are launching dozens of new EVs in the next couple of years have to meet. Tesla’s charging solution is slick and intergrated; the patchwork of rapid chargers, all with different models and sign-up requirements are poor competition currently.

And Tesla is not standing still with more chargers planned for 2020, though not sadly for North Norfolk yet. I can but hope…

The figures

I owned my Mercedes E Class estate for nine years and drove it 116,067 miles. I put 14,822 litres of diesel into the tank over that time at a total cost to me of £20,470 or thereabouts.

Much worse than that, though, is that over that time I contributed 39.7 tonnes of CO2 to the atmosphere. That is the main reason I wanted to make the switch away from fossil cars.

The Model X was more expensive than the equivalent diesel car (though it’s hard to compare, really, as there is no direct equivalent as any EV driver will tell you once they own one.)

More expensive to buy to much cheaper to run

But servicing and running costs will be considerable lower. The Mercedes, which, incidentally, was the most reliable I have ever owned, cost me about £700 a year in servicing costs excluding tyres and extraneous costs such as body repairs etc.

I would expect my Tesla to be a fraction of this as it doesn’t have oil or oil filters to change and has brakes which don’t wear out.

And running costs will be significantly reduced. I estimate it would cost me £6,134 for the electricity I will consume over the next nine years which means I will save over £14,000 on the running costs of the Merc.

Oh, and by the way, my electricity comes from Ecotricity and is entirely green.