Coding in the age of AI

Also: my love/hate relationship with BankID, Apple opening up the NFC chip for Vipps, EU AI act and how to design carbinterfaces

My love/hate relationship with BankID

I just bought a car. Do you care to guess how many BankId signings I needed to buy the car and apply for a loan?

  1. Sign contract

  2. Sign in to apply for a loan

  3. Sign in to SBL/Altinn

  4. (I also needed to log in to my bank and Lånekassen to find information for the application)

  5. Sign in to my page to see the loan offer

  6. Sign a confirmation of identification (ūüôą)

  7. Sign in to see the loan offer

  8. Sign loan offer

  9. Sign in to see the mortgage document

  10. Sign mortgage document

  11. Sign in to The Norwegian Public Roads Administration to change the owner

  12. Sign in to Digipost to see confirmation from The Norwegian Public Roads Administration

  13. Authenticate again to see the message of the owner change in Digipost

Yes, we are fortunate that all this can be performed digitally and that all these services talk together. (And yes, some of these sign-ins could have easily been avoided). But it poses a more significant question if Bankid should try to make multiple authentications easier or less strict. For long, I've been annoyed that SBL needs another BankID signing on loan applications, often right after you've just authenticated yourself with Bankid. I know this is because the process needs to be secure and robust customer authentication is needed, but imagine how much smoother this process could have been if Bankid was able to persist my authentication, and if I, for the next hour or so didn't have to go through the entire sign in process. BankId could then only ask me to confirm it's still me in the app, and not ask me to type my password 15 times.

Coding in the age of ai

Raido Pikkar, Chairman of Thorgate Group, writes in Shifter that:

"The best developers will be able to do most of the work thanks to artificial intelligence. Pure code writers will, in practice, become redundant." He continues: "Developers cannot limit themselves to being pure code writers. They must think more broadly and strategically about customer needs and business context, as well as place a greater emphasis on innovation and renewal."

There might be something in what Raido Pikkar says: Cornell University conducted a study using data from an online labor platform, challenging the notion that AI will only benefit human workers without replacing them. The research highlights instances where AI tools are already replacing certain jobs while boosting earnings in other sectors. 

The study focuses on language translation as a prime example, revealing a 30% decline in earnings as AI tools like ChatGPT and Bard outperform human translators. Additionally, the research suggests that fields like web development and machine learning are currently in a "honeymoon phase" of AI assistance before potential substitution. Notably, the study proposes that Machine Learning Engineers face a higher risk of being replaced than Software Engineers, citing the linear structure of ML code and the greater availability of publicly accessible ML models as contributing factors. The overall message discourages switching to machine learning as a safeguard against AI replacement, emphasizing the broader impact on various professions.

In my opinion, many developers already think more broadly and strategically about customer needs and business context (at least at Stacc and other companies I've worked with). Still, I think a good comparison here is Chess: almost thirty years ago, computers started to beat the best chess players, and people feared there was no reason to play Chess anymore. Yet Chess has never been more popular than it is today. AI has actually helped revitalize the game. 

James Somers has written a beautiful text about this and his feelings around coding as a craft in the age of AI: "The thing I'm relatively good at is knowing what's worth building, what users like, how to communicate both technically and humanely. A friend of mine has called this AI moment "the revenge of the so-so programmer." As coding per se begins to matter less, maybe softer skills will shine."

GPT-4 is impressive, but a layperson can't control it as a programmer can. Maybe the future of the programmer will be more like the quote attributed to Einstein: "If I were given one hour to solve a problem, I would spend 59 minutes defining it and one minute resolving it."

Evan Armstrong also writes about this in Automate the Simple Stuff With Stupid Tools: "Over the next few years, a decent percentage of thinking you do today will no longer be required (or compensated). This is the dumbest this tech will ever be. It is the least used, least deployed state of AI. From here, it only gets more sophisticated and more integrated into our lives."

EU AI act

Last week, European Union lawmakers secured a political deal on a risk-based framework for regulating artificial intelligence (AI). The negotiations spanned almost three days, resulting in a series of prohibitions for AI applications. The key points include:

  • Biometric Categorization:¬†Prohibition on using AI for biometric categorization systems involving sensitive characteristics such as political, religious, philosophical beliefs, sexual orientation, and race.

  • Facial Recognition Databases:¬†Total prohibition on untargeted scraping of facial images from the internet or CCTV footage for the creation of facial recognition databases.

  • Emotion Recognition:¬†Ban on applying emotion recognition in workplaces and educational institutions.

  • Social Scoring:¬†Prohibition on social scoring based on social behavior or personal characteristics using AI.

  • Manipulation of Human Behavior:¬†AI systems cannot manipulate human behavior to circumvent free will.

  • Exploitation of Vulnerabilities:¬†Prohibition on using AI to exploit the vulnerabilities of individuals based on factors such as age, disability, and social or economic situation.

The EU lawmakers seem to have struck a balance between innovation and safety. Still, some civil society groups express skepticism, particularly regarding limitations on state agencies' use of biometric identification technologies. The formal adoption of the law awaits votes in the European Parliament and the Council.

Apple opening its NFC for Vipps?

Apple might be opening up its NFC chip to third parties, which means iPhone users could soon use other digital wallets, including Vipps! Vipps CEO Rune Garborg is thrilled but cautious, waiting for official EU news. Vipps has, over the last few years, criticized Apple's NFC monopoly, preventing customers at DNB and others from using Apple Pay. Now, it seems Vipps might be winning as Apple proposes a solution to the EU amid a competition case. The twist comes with a potential fine for Apple, pushing for fair competition. It will be interesting to see how this unfolds!

How to design car interfaces?

We jokingly say that we in Stacc work with "Bank og Bil (cars)." This is even more true now that we are a full-service provider in the asset finance industry. Personally, I've always found the UIs for car interfaces fascinating (and often bad). For this reason, I was thrilled when I saw The Turn Signal and its blog post on The 8 Guidelines for Good Automotive UX Design. Most of these guidelines can also be transferred to other UX designs. If you are interested in the topic, I also recommend these interesting blog posts: